Despite our many disagreements, I wanna go ahead and say that I believe Steven did a great job presenting his points, and that I wish him all the best with his endeavors for applying pressures where needed - Please be kind and respectful to him and his community when you visit his channel and watch his original video here: th-cam.com/video/tjSxFAGP9Ss/w-d-xo.html&ab_channel=StevenZapataArt Also, shout-outs to Jazza for the hilarious Fiver video on his channel which can be found here: th-cam.com/video/5GO2xKmZsVo/w-d-xo.html&ab_channel=Jazza If you're interested in creating fun art and installing Stable Diffusion locally on your computer, you can checkout my video playlist explaining how to get started here: th-cam.com/video/6MeJKnbv1ts/w-d-xo.html&ab_channel=RoyalSkies As always, hope you have a Fantastic day, and I'll see ya around -
I think you should have an actual debate with this person instead of making a "reaction video" (I'm not being sarcastic or condescending here I just don't know what to call it), as this method gives you the last word on pretty much every point making this all sound one sided. While I agree with a lot of what you said and you are still missing and/or didn't address some really important parts. Many of which come from your own perspective as a successful artist.
@@ScottSpadea You made several interesting points, but I'll just raise one: The AI is not intelligent, it is a flawed program written by humans who admit they don't understand how the program produces its results. No program ever written is perfect. No computer part is made without a mistake. Add these up, and throw in a data-set of information from people who hate, despise, refute, ridicule, belittle, and reject real art for their own personal pleasure, and the AI is going to spit that out, more or less, in every image. The problem is the human condition, not the technology. The motto of the computer scientist is this: "Garbage in, garbage out". Let that sink in before you decide to take a trip in an AI vehicle. Love the technology, be cautious of the people who produce it.
Wow, I don't know where you got the link from, but I did send it to you, glad and surprised you could take it that seriously - from 90 seconds to 90 minutes! Is that a record? I actually don't think you addressed the points raised, although you went through them, so I am still not sure what to make of his arguments, but at least you've clarified your position. I'm happy to go through all the points, but not sure what format to use. I believe you both have more in common than you think, but it's like describing different parts of the same elephant. You both want the elephant just don't agree with the part the other person is holding on to. Personally, I'm not into AI nor Art, but what I understand about both is that the programs are blurring the lines between copyright law, intellectual property, and ownership of a work. It also makes your definition of art - which I accept - difficult to apply to a program, which is what AI is. So I don't have a problem with the position you hold, and it is probably a positive result for the 20-30 years AI researchers received mockery and ridicule for getting no-where. Whose laughing now? (ssshhh.. I still don't think it is intelligent...😉) That said, some of the responses didn't seem to touch on the thrust of his point and I'm not sure if I missed the actual points because I'm not an artist, or that his argument wasn't addressing the actual issues. And that's what I'd like to resolve, if possible.
I swear, more half of Steven's video is spinning up an "Us vs Them" scenario using the most emotionally charged terms and constant assumptions of malice. Its so frustrating trying to see his point of view when the entire ramp up to his point is inflammatory fluff, the crux is ignored for a single aspect, and the payoff is condescension and calls to action. Distilling some of these talking points could be summarized as "you want X? you don't want X. You're bad for wanting X" with only hyperbolic rabbit-chasing to back it up. Royal, you did a fantastic job pulling the most complete interpretation of his points out and addressing them fairly. And for all those years you spent learning 3d modeling, the grind of working out code problems, and dealing with arrogant cockends who seem to get all their sense of self-worth from internet forums, I can't tell you how much time your guides have saved me. Whatever your process is for trimming out unnecessary information and compacting your presentation is, it's absolutely worth it from a 3D hobbyist's perspective.
Nope, the video from Solar Sands is under viewed, everyone should watch that before making a final conclusion on it, it is not just simply about new technology and full stop, it is very complicated
Watched the whole thing, and I've been thinking about it. I think i decided that no matter what my opinion is, I respect you greatly for doing research and keeping a level mind, and generally not being a jerk either for or against AI art. I'm not sure yet if i agree with you on everything, but I definitely respect you. Thx for putting in the time and effort
I remember watching part of this video some months ago. I agreed with some of his points, but i didnt finish it because the narrative is clearly and sharply based in fear of the unknown. I sure hope ai isn't abused by or for corporate profit margins, but i definitely think everybody artist or not would enjoy or benefit greatly from these types of advancements. I'm not an advanced artist, and i havent spent much time or have time currently to learn more on blender, but i can see indie game devs getting a massive productivity boost from ai advancements. Just like how 3d modeling software advanced in the last 40 years.
I'm glad Shadiversity recommended me to head over here. Suggested a friend to watch your playlist on Stablediffusion and appreciated it. I'll give a look at that playlist as well since I think it can probably help me visualize my ideas from descriptions of my characters and create something cooler than what my words describe. Thank you for making this response video.
Technological progress will never be stopped. We know this simple fact that has defined humanity for years. The printing press made it far easier to mass distribute writing, the typewriter revolutionized writing large amounts of words like novels, the automobile revolutionized how we get around, and the personal computer changed how we live our lives every day. And now we have AI which is poised to revolutionize how we can make art
As a computer engineer that has also worked on AI, thank you for clearing up misinformation. I dislike elitist people like that dude, half of his points were "I'm a real artist" basically, not to mention that people like him forget that AI has been available for years already and in far greater and more useful fields than drawings, like for example in medicine, but you don't see doctors protesting, they actually love its help.
a calm and direct response towards what I can only describe as pretentious, presumptuous and sophistic rhetoric which sadly a lot of artist are very easy to fall prey to
This is sorta out of context, but 35:39 this story really hits me. I've had the same experience reading trough programming fourms and people just refusing to answer (most recently with linux). When I was starting out in blender and I had the dumbest of issues, your channel was so informative and so straight to the point that not only did I fix my problems, but I binge watched the rest of your videos and learned so much from it. Things that I would have waisted hours of my life because I forgot to check a box were now something I knew how to fix and not only that, if I forgot it or if it had some small differnces, I always had the option to check your channel and watch the super quick video again... IDK if you get this often, but thanks for steping out and being the opposite of the fourm guy in the 35:39 story.
I remember as a kid I tried to make Minecraft mods and people would just say "Learn Java" instead of helping with the issue. When I got older I relized they are just assholes who know enough programming to use it but not to teach it. I'm pretty good at programming now but honestly you could learn a decade of my experience into one year nowadays and I'm happy about that.
i wonder how that guy feels about schools; i bed he'd make a greeeeaaaat teacher. student: Mr.shitbag, how do i do #5? Mr.Shitbag: that's Professor shitbag to you, and if you wish to know how to do it, you will have to figure it out yourself, or go to a 4 year college like i did, where a professor taught me everything i know.
The story at the timestamp you mentioned. It hits me too. so we're in agreement on that point. But when we're talking about AI-ART, most artists don't argue "it took me years to achieve this, its not fair you can do it in a click of a button" They dont, not in terms of Ai-art. Because "at the touch of the button" has been there for years, in things like 3D and lighting. You don't need to understand lighting, you just need to place it etc. When we're talking about Ai-Art, the argument is "it took me years to achieve this, and its not fair that you stole this work from me to / used my work to train your Ai and call it yours" Thats why most artists are ok with Porgrams like Adobe's Firefly, that uses only paid-for/licensed images to train their Ai.
I am having almost the same issue on my project. I actually need his lua script on how he made the enemy AI work (making the enemy, smartly, jump out of the water towards the land and hunt you down especially if you're not on the water. They have no issues jumping down the water though.), but to no avail. I even warned that there will be jank on my enemy since I couldn't figure out how to make that kind of script nor get to know the secrets on how he made it. I keep asking this for like 10 times already only to ending up with a straight "no" to the point of me going to the extreme that I am already harassing him that I really have to stop nagging about it because I am getting too far on my methods. The TL;DR is. He's not willing to give the script nor tell me the secrets, not even a single tip or, hint on how he did it to the point of me getting a bit pissed off and going a little bit too far.
I agree with you that the narrative AI vs. artists doesn't help at all. But I don't think that's the point. The narrative he seems to make in the video is artists vs. the companies selling AI and I think that has a point. Because Google won't come to you, asking for your images to automate the process for you. They will either charge you or automate it for themselves. I know this sounds stupid because especially Google is mostly a service provider currently. But the difference is that while they only provide service now, they can provide the product (art) themselves later because of the AI training you agreed on. Because with your art, you are giving them your personal experience for free to feed their algorithms. You experience is the product here and you are giving it away for free or far too less, I'm telling you. Of course those companies won't start doing that tomorrow but they will get there. Now they are all still improving the algorithms. So it makes more sense to allow others using it as long as they feeding it with data. But as soon as they cut the line, your career as artist is probably over. With more and more digitalizing, I don't think physical art or "human art" will make much money in absolute numbers either. Sure, it will work for a small amount of artists being kind of celebrities because that's how premium branding works. But most people will take AI generated images over manually crafted artworks anytime because the density of content is higher, details will be finer, resolution bigger and cost much lower.
@TobiasFrich "They [Google] will either charge you or automate it for themselves. " I think you're on to something, but not quite there. The point of Google's ToS is to transfer your copyright to them so Google controls the distribution and sale of your work, not you. And this means any company who copies your work into an AI dataset is violating Google's copyright because Google doesn't allow it. And this applies to all companies that require you transfer copyright to them. They want control over your material for their profit, not other companies, and certainly not for you. So whether or not you signed away your copyrights, the work is still under copyright law, and has a legal copyright holder that can determine the conditions for copying 'their work'. They will, of course, freely allow anyone to copy your work, for a price. And that is where the dataset company does the double take - "we're a non-profit, for research entity and therefore not-for-profit and can legal copy your works under fair use, without payment." After which, the dataset is moved to a for profit company, completely legally which they describe as a capped entity (or whatever). So, now Google's attempt to profit from your work has been wiped away by these entities because they can 'sell' your work to other companies at a 'discount' compared to Google and every other company who got your copyright. ooof!
@@petergostelow To elaborate on a key point: They have rights if you upload it to a Google service, not if it's automatically added to their algorithm. That's the key difference. If I make a website and Google takes the images off it, Google doesn't own the rights to that image. They even say it themselves: "Google does not own the images found via Google Search. The "Usage rights" Search tool is provided to help you find images which may be suitable for your use." I'd post the link to the article if TH-cam didn't filter my original comment with the link as spam, haha! I'm sure if you search the quoted text in Google, you'll find the article, though. Google has absolutely zero control over images in the search results unless you manually upload them to one of their services!
@@mbishop3d832 You're right, although I wasn't referring to the search engine, neither was the video. Somebody always has copyright, even if it is public domain. So anyone who randomly copies an image could well be violating copyright law, whether from the search engine's results, or your website.
@@mbishop3d832 Yes, it was with reference to 7:28 'counterpoint to stolen art'. I am just pointing out that all works have a copyright attached to them, regardless of the Terms of Service because the ToS does not remove the copyright, it merely transfers the rights. Thus, the ToS is not a counter the theft claim. It's true they're not violating your copyright because you no longer have one, but somebody does, and that's why companies have created the work-a-round capped entity. The 2nd fact is, that the companies have spent large sums to create a specific economic entity to download, should alert you to the probability their lawyers know something we don't. So the ToS is irrelevant to the theft claim, imho. Did I miss something?
I want to point out that - "Web scraping is completely legal if you scrape data publicly available on the internet." meaning they don't even have to get you through the terms of services of other platforms, and then buy the data second hand, or second rate. They can kick it up, and just get your data directly from the source no middle man required. It's not theft if it's legal. You may not agree with the law. But when you use legal terms such as theft, the law has to be taken into consideration. And considering the laborious processes surrounding AI Art, there's a strong case it falls under fair use. Which would grant a level of legal immunity, if granted by the courts. If your art is private, it's not protected from viewing. AI does not cut up and rearange your image, it views said image. But it doesn't use the image itself. It use the data it receives from said image. And creates based on its programing it's own image. The issue lies with the multitude of images. No 1 image makes a single AI image. Millions upon Millions images (data), related to the prompt are used to create a new image. In that sense, it can very easily be argued as transformative. Because 1 image does not lead to another, it's entire dataset does. Meaning it's essentially drawing "inspiration" from all corners of it's databanks. Thus making it impossible to draw a directly line from 1 image to another. Unless a separate method of AI Image Creaton is used.
12:07 yeah this is a pretty good summary of 95% of all internet conflict. The "us vs them" mentality is simultaneously very enticing and very destructive, being a carryover from the stone ages and all. Zenata is purposefully getting the audience riled up and angry because anger bypasses rational thinking, and makes it far easier to draw in onlookers that would otherwise be neutral. It's rather insidious
I am an artist and have messed with the AI programs, I can say that the machine by itself will not be able to replace an artist unless (ironically) and artist works the machine. Not to be rude to Mr. Zapata, but though his argument is good, there is too much emotion and fear-mongering covering his points up. Also, you could change the topic of "AI art" to something like "game engines" and the argument would sound similar. Though I do know bad people are going to abuse the generators, I don't think scaring people, especially artists, from using the programs is going to resolve the problem, it actually might make it worse. Edit: NVM Zapata completely lost me on the second half when his arguement became an _"arguement"_ ... well, back to drawing and rendering with Stable Diffusion at the same time. Good Vid Royal Skies, glad Shadivesity directed me to ya :)
About the feeling argument at 1:02:00 There are a lot of people that enjoy growing vegetables in their garden. By the "artist" logic we shouldn't mass produce vegetables not to hurt their feelings
i think his point about your art no longer mattering is more a point about making it even harder for artists and AI generator users to turn their passion into a living, which is a valid point since there'll be a billion more people flooding the market with their generated images
Ehh, as if there aren't a billion starving artists competing against each other already. This just proves even further that without sheer luck, art is not a viable career path.
@@DonVigaDeFierro oh it's definitely not a case of just luck currently, if literally any potato can just type "gimme pretty picture" into a textbox it may well become that way though
@@tieguaili3d298 While "gimme pretty picture" is pretty funny, I'm genuinely tempted to put that into stable diffusion, I'm curious as to what will happen!
ANYONE can draw. Not everyone can draw well enough, or train to get the aesthetic sense that people will PAY for. A billion people making shit are still people making shit. AI is not capable of, and in the near future will not be capable of, producing unaided to a specific brief. Now, if one's entire living is based around making pretties as fine art, well...that might be a tough sell, much as it is a niche market for artisinal glassblowers (which still exist, despite glass manufacturing being automated to a large extent).
they're not automating the boring shit though, if they were they'd be automating retopo, UVs, rigging, technical bullshit that takes forever and no one wants to do themselves cause they're rather be sculpting painting and animating, but instead the AI people are automating the fun creation parts not the boring technical prep parts
THANK YOU! This is what those ai programmers don't really understand. If there was an AI for retopology, that would be amazingly useful. Currently AI programmers are doing the wrong thing, they should really consider making things like you mentioned (AI powered retopo/UV plugins) instead of sapping the fun out of creating stuff.
@@quadrantthree static image generation was easy pickings for programmers. There's already a lot of algorithm based retopo and uv based automation. But it's a lot harder for an AI to understand it for example texel density and specific layouts in uvs are often frankly more subjective to the project or task. Than feeding a bunch of results into a model for replication since ideally all results fed into the AI were generally the best examples of what it should do. When all the information is selected by us then any result given to us by stable diffusion will be good basically. The moment you ask it to do something it hasn't quite seen you'll begin hitting roadblocks. This is why stable diffusion video is trippy currently. It'll take 8 more years to fully integrate AI into 3D and 20 more years to integrate AI into visual effects and engine programming/coding. And that's if everything goes well
You have no clue what is hard for AI to do. What you think AI scientists and engineers should do is irrelevant, because you don't know shit about the field. We're building intelligent system from the ground up. This type of technology, text-to-image models, weren't created with artists in mind - they're all subsets of natural language programming, computer vision, and, with diffusion models, thermodynamics. The fact artists got hit first is just an unfortunate result of what was discovered first. Artists may have always been the first to be hit, just because the visual components necessary for art is just in line with computer vision and image generation. Unlucky. But dont tell me and my colleagues "what we should be working on". We've built this field from ground 0. We discover things over time and automation follows from our findings.
@@leamJG how bad does your butthurt have to be to go off on a rant like that because someone pointed out that your discoveries are of no use to anyone and you'd have less people trying to get your shit shut down if you looked outside your little bubble?
@@tieguaili3d298 discoveries are of no use to anyone? You know AI is used in cancer research and in the health field, right? We can identify cancerous tumors better than doctors can now. We can accurately predict and simulate epidemics. Our work on neural networks have helped neuroscientists learn more about the human brain. Get your head out of your ass.
“You want it to matter you told your story.” I mean, what? I don’t sell any AI art, I don’t post anything, I just show stuff to friends and family who are interested. The fantasy warrior with armor that looks like a jester with flaming swords matters to me because that image was in my head for years. I could care less whether other people validate me, or if they think it’s stupid, it matters to me that I can finally see it. Those who think that humanity now has that ability have their priorities completely out of whack, they care more about praise and validation than actual imagination. (Edited because I duplicated a words)
I think what ticked artists off the most was the creator of Stable Diffusions comments on Artists "My message to them would be, 'illustration design jobs are very tedious'. It's not about being artistic, you are a tool". That rubbed everyone the long way. They make it seem like it's tedious when every brush stroke is a blast and instrumental to an artist's well being. It's not about the final picture, it's the journey there. The most popular art in history is unfinished. This is going to replace FIVR artists if anything first.
I actually fall into Royal Skies' category of artist... I enjoy the end product, but find the process very frustrating and tedious. Different artists find different aspects of art fulfilling. Mine is in getting ideas out of my head, and onto a computer screen (or onto paper, as I like to sketch ideas quickly sometimes.) I'm all for AI art generation tools. I specifically like the image2image function that can take my original concept art, and then iterate on it, adding details and such that might inspire me to add other things and change it around. I also like the functions of inpainting and outpainting to fine-tune certain parts of the image, allowing far more control of the results. These tools have re-invigorated my desire to work on concept art, whereas before, due to the tedium and frustration, I had been suffice to just imagine them for quite some time now. (as a writer, I write descriptions of my characters, and generally draw a concept sketch of each one for personal reference and inspiration for writing, but AI tools have opened up the idea of illustrating far more scenes and locations and characters in the book.)
Good points made here, but I want to say that really this is just the end result of the real problem that's been plaguing users of software for over a decade now: there is no functional consent on the internet. ToS isn't really a choice; every piece of software and every online platform collects data, and there's nothing you can do about it unless you just don't use that app/platform. Often, the ToS isn't even accurate to what the company wants to do, rather they state everything they *could* do, just so they don't have to make you sign a new ToS every time they change their business strategy. It's also hard to justify it as a choice when these companies actively shoot down open source or similar alternatives, either through acquisition or litigation; there is no guaranteed alternative, so there's no real choice here. Additionally, there is no real safeguard preventing theft on the internet. If you can see the content, a bot can see it too; if you can screencap the page, a bot can too. You can lock it behind a paywall, and that's likely your best bet nowadays, but that limits your reach. As a hobby-ish artist (I also use it for professional purposes but am not a commission or salary artist), I just accepted that my art is in the public domain and can be captured by anyone or anything for any reason and at any time. This makes me apprehensive to post a lot of my stuff, however, especially the stuff I might use in an IP later on. Personally, I'd feel much better if the ToS didn't allow for the platform to share my data without some additional agreement that isn't mandatory for using the platform. Since I can't stop users or bots from collecting it, I'd at least like to have my content and info safe from the platform or software itself. I think we've reached a tipping point with privacy concerns, so as long as people keep up the pressure I could see some beneficial changes coming to platform policies in the future. It's not just the images you post, but the information about you that you do not know they collect (I could have agreed to share my name, for example, but it doesn't mean I consent to you using that to find my email from a database), so it's an overall privacy issue imo. As for AI, though, I understand both sides. I prefer the open-source approach because it allows the AI to be art in itself after users are able to refine it for a given purpose; stable diffusion I think will go the path of Blender and surpass the paid alternatives. Also, most people who use open-source software already follow unspoken rules about plagiarism because there are no safeguards whatsoever in those programs. For the paid AI like Dall-E and Midjourney, I understand why they need to monetize it; the program is the product of their work, so they should be compensated, and I understand preloading a dataset because it's a core part of the convenience that makes up their competitive edge. You can't just remove all the training data and have users go hunting for thousands of images just to use their product. I understand the artists too; it feels wrong to have your work used to inspire a mass-production capable art machine. I'd also argue that this can extend your reach; personally I would feel flattered that people are trying to emulate my style, and so long as they share my name alongside the AI art, it'd expand my viewership to niches and groups I don't or can't fill with my current portfolio and skills. In the end though, the ethics of AI art are entirely in the hands of its users. Plagiarists will plagiarize, those using it for inspiration will continue to do so, those using it as a tracing sheet for learning will continue to do so. It's the responsibility of the community to hold each other accountable, and I just fear the legal system getting involved because they will inevitably do something stupid like more heavy-handed requirements for image licensing
Art Generators: I make this cool image for my twitch subs of my dog who passed away. Artist: you hate us and want to enslave us. You want to make us less than you. I'm glad more artists don't have thay guy's mindset
Is it fair for me to draw then use AI to enhance or change the drawing somewhat in order to quickly draw something from my drawing? I want to start using this to help me draw or add color, since i've never been able to learn it properly. Is that fair? Using it to help my drawings? Making make some money after all these years?
Probably but the image generators were created using misappropriated / unlicensed assets of other creators. Using software based on misappropriated assets seems unethical to me. The software developers did not even bother trying to make it ethically.
Don't let the people fear-mongering about this technology scare you. The companies who trained the AI did not "misappropriate" anyone's art. If you are not comfortable using someone else's art for reference, then you shouldn't use these technologies, but if you are comfortable with that, then I don't really see the difference. Whether it's you personally viewing the copyrighted images in order to produce new images, or the AI training on the images to produce new images, I don't really see any difference ethically speaking, so long as the images produced are genuinely unique.
@@mf-- Flat Earther tier logic. Nothing was 'misappropriated' lol, unless you think looking at reference images is also an example of 'misappropriation'
Thank you from the bottom of my heart for this video. People like Steven only cement my misanthropy against society, especially artists like him. We have this problem with his kind in the gaming industry and it is one of the reasons I despise the industry's freelance scene and the so-called industry "professionals" in the AAA scene from the likes of IGDA and it's chapters, Kotaku, Polygon, the jokes that are the AAA companies. They are clowns and often abusive to anyone that does not share their zealous mentality complex. Some of these freaks will try to cancel and destroy our lives for simply not liking our faces, hence their Cancel Culture. This Art Renaissance has honestly rescued the free peoples of the gaming world from these creeps who hijacked and occupied it in the first place in the 2010s (or arguably the late 2000s). I know any actual artists who care about creative freedom and values free enterprise will have nearly no problem with this breakthrough. It is only the elitists and the zealous prejudicial ones that are envious and afraid of this change, because they know they are a major key threat, and it affects them. The evil is shouting it's way to it's grave. It is time for them to pack their stuff and leave us alone, so we can truly thrive as a free world again.
I'm sorry but I strongly disagree with your counterpoint to AI not replacing artists. I understand that you might think the artist will just be the people using the AI as a tool but it's not designed that way. The AI is developed by computer scientists who couldn't care less whether artists have an upper hand in using this software. They design the software to sell to masses, not artists. So essentially the goal is too make the best images with the least amount of knowledge or experience because that is the optimal product in this line-up. That is what sells... Nobody wants to pay an artist to fill out a form of text and click "generate". People just want to have beautiful images and the thing AI allows those companies to do is to make creating the images a personal experience. This is something artists can not compete with however you try to spin it. Because art is subjective... you can not order an artwork from an artist and expect a perfect result which matches your exact idea/concept/though. But you can order that from an AI because an AI does not have own subjective preferences. The bigger the training set is the lower its bias will be overall. So as an artist you will compete with your artwork against (essentially) your customers artwork... I don't think, I have to tell you that you'll loose.
People do not “just want to have beautiful images,” they need specific images to a brief. I’m not sure if you work in the field, but if you did you should know that “beautiful images” is just one part of a very large job as a professional artist.
@@done.6191 This was a simplification. What I mean is that most people who would request an artist, they don't really know the exact requirements what they looking for. Even if they know, it is pretty difficult to express it. But in the end whether they like the result or not is subjective. So you as an artist do not have any edge over artificial algorithms which can generate piece by piece until the customer is fully satisfied. Because this eliminates the loss of information in communication between artist and customer. It is also technically possible to adjust the algorithm dynamically to any customers individual preferences during usage. That means the amount of work they need to put into filling out forms can reduce over time to get requested results. There's no way any artist using the same tool could achieve that while meeting multiple customers subjective preferences. The tool can also be marketed as providing any customer an individual experience and you get the advantage that any image or other piece of art produced by it will have a different status. Because the customer will not just have paid for it but they will produced it themselves (in a way). So the subjective value of every piece created by themselves will be higher than anything in contract with an artist using the same tool. So in the end these algorithms will change the art industry completely, thinning out professional artist as job itself. I don't think it will make physical art by professionals obsolete and some very popular artist will likely stay relevant. But I don't see high chances for artists currently contributing in video games, web design and other low budget contracts. Because using the AI will be more economical and people won't use an artist as operator producing higher cost.
@@TheJackiMonster I have a huge edge over AI, as I can distill extremely complex briefs in a very short time. And AI cannot complete this task, as the time for this to happen: "artificial algorithms which can generate piece by piece until the customer is fully satisfied" approaches infinity. It isn't that it is HARD for the AI to do, it is that it CANNOT DO IT AT ALL. Until an AI can process language as a human, IE essentially until AI can BE human in all but flesh, it will fail at complex tasks requiring this fuzzy thinking and free associations that humans can do naturally, and do very well with the right practice and training. What AI will do is accelerate the pipeline, as it has been doing for a decade now, by reducing the step-count for mundane tasks. It may be that one day, every artist will become a part of an AI/Human team, with the AI handling grunt work and the Human functioning to put all the pieces together, ensure visual cohesion, produce concepts the AI can't do well, and overall art-direct the bot. Ideally, the bot and human will become like coworkers, and learn from each other, developing a style and technique unique to that specific pairing. Also, game and film art is probably one of the HIGHEST budget contracts, not the lowest. It requires a constellation of skillsets, all meshing together in a complex technical and artistic environment, to do properly, and as such are generally well paid. True, they don't get paid $90 million for a picture of a dude staring at a pool, but that's because the fine art market is a huge scam/tax avoidance scheme. The tools will improve, and yes for some tasks "I need a red coffee mug with a pencil in it on the table," they will be automatic. The rest will either be the sole domain of human artists, or will use Human/AI teaming to get the job done faster.
@@done.6191 If you really think you have an edge over AI because you understand human language, you seem to have failed to understand my point. You can completely eliminate the latency and potential information loss of communication by using AI tools. Because a customer does not need to speak with any artist at all to achieve their goal. They can simply tell the machine what to do. But besides there is actually a lot of research going on for algorithms to process and interpret the human language. So it's not a stopping point for anything in this process either. However it is far more likely that any human will just learn to provide a machine the necessary pieces of information in understandable terms for the machine. Effectively that's what computer scientists do day by day. So don't tell me: "No, only an artist can do this part!" or "Artists can do this part better!" Maybe in some cases experience in art will help quite a bit to produce targeted results with those algorithms. But as I stated before: Computer scientists will develop those algorithms. They will implement the user interface and all options to configure the usable process. They will design it in a way you don't need to have any experience in art at all because that is what sells. So no, there is no edge over AI which matters here. Because artists won't have any edge over other human beings without their experience anymore despite the produced results will be art. That is my argument. I'm not saying the machine will replace you as an artist but it will kick you out of business. Isn't that kind of obvious looking at the descriptions of those tools? Guess why I'm advocating for artists claiming their copyright here? If you think this discussion is about human vs. machine, you are wrong. It's about human vs. human because artists are certainly a cost in production quite a few people like to get rid of with automation if possible. I mean there is a whole sector of research about procedural textures, photogrammetry and others which partially allow that already.
@@TheJackiMonster “you can eliminate the latency using AI tools.” “You can recombobulate the fruction with an XG modulation.” These two statements are essentially equivalent. There has been ongoing research into human speech recognition for decades. It is still not even close. AI won’t kick me out of business anytime soon, until it kicks EVERYONE out of business. Just saying stuff doesn’t make the words true. Scientists are working on these algos, as they have for decades, and they’ll continue to try to bottle lightning for as long as people are willing to give them resources to do so. I’ve understood everything you said, I just don’t think you are right. Computer scientists are not artists. As such, they won’t be able to surpass the best human artists (and in some tasks, even the worst ones, ever, until they create synthetic life. At that point, all bets are off. There is so much hand-waving in your view here that I would mistake it for the Queen.
9:21. With due respect, I have to make a solid objection to your counter. I'm for my fair share of usage of AI elements, let's get that 100% clear, but this argument REEKS and I wanna call it out hardcore. EULA's are absolutely legally binding, however, to immediately claim that somewhere along the line that all this data becomes EULA compliant is absolutely absurd. One of the biggest platforms, one of the top prompts used in most AI art even, revolves around a website with the following statement in the Terms and Services: "As between you and Epic, you will retain ownership of all original text, images, videos, messages, comments, ratings, reviews and other original content you provide on or through the Site, including Digital Products and descriptions of your Digital Products and Hard Products (collectively, “Your Content”), and all intellectual property rights in Your Content.". ArtStation itself has it clearly written in its legal dialogue that all rights are pertained as originally distributed prior to its uploading on the website. If somewhere along that line you might use a piece of software with some sort of EULA clause that might be applicable, say perhaps an Adobe product, it would be through Adobe you'd have to get those rights and NOT through ArtStation. ArtStation has its fair share of traditional media content on it as well, which could easily make the process from paper to ArtStation short enough that the rights would inarguably be only yours. The REAL reason most of these datasets are "legal" actually comes down to Fair Use, but that's an important distinction to make. Many of these datasets, such as Stable Diffusion's checkpoints, run under GNU GPL licenses, and therefore can be distributed under educational licensing. Stable Diffusion also leaves this little nugget in their license you ought to read "no event and under no legal theory, whether in tort (including negligence), contract, or otherwise, unless required by applicable law (such as deliberate and grossly negligent acts) or agreed to in writing, shall any Contributor be liable to You for damages, including any direct, indirect, special, incidental, or consequential damages of any character arising as a result of this License or out of the use or inability to use the Model and the Complementary Material (including but not limited to damages for loss of goodwill, work stoppage, computer failure or malfunction, or any and all other commercial damages or losses), even if such Contributor has been advised of the possibility of such damages. ". In lame man's terms, all this license says is "Oh, yeah, you can use it commercially, but if someone comes to sue you or people start hating you for using it we're not responsible." DALL-E's license says something very similar: "No Infringing or Harmful Use. You must comply with our Content Policy, and you may not use DALL·E in a way that may harm a person or infringe their rights. For example, you may not submit Uploads for which you don’t have the necessary rights, images of people without their consent, or Prompts intended to generate harmful or illegal images. We may delete Prompts and Uploads, or suspend or ban your account for any violations. You may not seek to reverse engineer DALL·E, use DALL·E to attempt to build a competitive product or service, or otherwise infringe our rights. You will indemnify us for your use of DALL·E as outlined in our Terms of Use." To be completely frank with you, that argument you put up there is from my understanding as someone who's been working all semester on a project discussing the legality of AI art, very, very wrong. Also, one last little snip at ya if you don't mind too much, while the AI might not carry the images itself with it, there's rules of consent around even allowing an image to be studied for reproduction. Unlike humans, algorithm generations do NOT fall under Fair Use, so the only argument that allows this practice to continue is the same one that got this AI here in the first place: this is a technological study. If you want a legally solid model, it wouldn't be impossible, but if you can type "Jim Bob ArtStation" into the model and you get a good representation of Jim Bob's work, EVEN IF IT GOT THE WORK FROM ELSEWHERE LIKE GOOGLE, so long as Jim Bob wasn't the one that put it elsewhere, he's well within his rights to mark a line in the sand in most cases. From there the arguments branch out into misleading Terms of Service, the ethics of Google's automatic image-grab algorithm, yadda yadda yadda. The point is that the line isn't as simple as "You signed up for this." I like these AI, I like technology as a whole and I think it's a fantastic thing as someone who struggles with dysgraphia and fine motor skills in general that there's a place for people like me to generate imagery we can use as inspiration for our own works. That being said, this model doesn't work without those images, even if it doesn't carry them at the end, so at best commercial usage of an unedited AI prompt I'd wager is "legally dark grey". Love your stuff, dude, and there's no bad blood on my end, I just want to make it clear that this argument ain't it, chief.
I feel like the potential to use the technology irresponsibly isn’t thoroughly addressed in these types of discussions. I’d personally find it concerning if 5-10 years from now AI art could imitate my art and pass it off as something drawn by me. I wouldn’t be able to prove otherwise and the soul and identity of the artist tied to the work would be forfeit. I’m reminded of the story of how not even a week after Kim Jung Gi passed away, his art was fed into an AI model and his style which he spent decades honing, became accessible via text prompt. I think as time goes on AI art as well as human generated art will begin to value showcasing the transparent process of creating the artwork. It’s naive to think that everything we do creatively can’t eventually be replicated, so is the value of the work we currently do subject to change as we become “obsolete”? If certain aspects of the process of making art feels monotonous, if it feels like work and isn’t fun or doesn’t leave you feeling fulfilled to see how you’ve improved as an artist. I don’t mind that you’d rather have a program do it for you. That’s entirely up to the individual. I’d be happy to have AI UV unwrap a model, or skin weights, or hook up shader nodes. But dedicated creators deserve to have their art identity protected and not emulated without express consent. Most tech companies are keen to protect their products from unlicensed usage. I can’t make an art asset in their software and legally sell it without owning a commercial license of the software. It’s a transaction at the end of the day. In the future, I hope artists can leverage their products in the same way before being stored in an AI database. As long as AI is done responsibly, I’m fine with it.
Beautiful comment, Austin. I've also heard the point made about artists providing footage of themselves creating the artwork to prove it's actually human-made. This would solve a lot of problems around authenticity. The question down the line is...will clients and fans pay for authentic artist created artwork when AI can create nearly identical art basically for free? I see a bleak future when artists can only make money by 1) selling art to their already established fans, or 2) targeting high price buyers who want that most novel and chic collectible...the human made artwork. If you look at internet-age revolutions like file sharing (Torrents etc), digital music downloads, pirating...you won't see much integrity. I'm afraid AI art will be the same. Fiction, Art, Music and eventually Animation will be AI opensource free-for-alls. Artists who devote their lives and talents will be reduced to referencing data points for AI prompt jockies who profit by cheaply and quickly making endless iterations which can match any art style/content. There's no way for artists to opt out of being mentioned in AI prompts that I know of. I read that there was a move to make this happen by a group of concerned artists but even if the AI opensource companies agreed to this, who actually thinks AI users won't get around that?
First of all if you shared your art online then it's already out there, people won't need an AI to copy your style, just some technical painting skills to make a reproduction. And if someone claims their art as if it was made by you and try to monetize it off your brand it's a crime no matter if an AI was used to produce it (and i don't count artist's name used in prompt because no one would pay for art because someone's name was used as a prompt only if marketed to be made by certain artist or because of the nature of final work, which is only partly inspired by a certain name in a prompt, no more than any of your original art was inspired by every art piece you've ever saw). So it's really not an argument against AI. If someone does it you can persecute it however you want, leave a disclaimer that your authentic art can only be found on your distributing places (site, social media or etc.) and anyone selling it otherwise is scam. If you wanna leverage in court that your "artistic style" is the same tangible concept as a source code of proprietary software and anyone using it without your concent is violating your copyright, then good luck to you and an art community as a whole. As music industry copyright bullshit clearly showed already, you won't be able to post shit anymore because you'll inevitably infringe on someone's style sooner or later and a swarm of copyright trolls will take any money you'll try to make using your art. It might be disheartening that you had to spend a lot of time and sweat to learn to do things that people can just accomplish with a "text prompt", but people care about it as much as any other automation field, so you have three options: be a luddite and reject any technological progress at all, be a hypocrite and care only about machines replacing exactly you or be appreciative of wonderful possibilities that new technology will present us and embrace change by changing yourself. Eventually people will have to fall in either of these categories so you must decide for yourself on which side you have more to gain for. There will be communities and marketplace for AI rejectors for sure but will you be successful in those is a big question. Rising against AI might sound like great idea but it will ultimately fail as most of luddite movements before, and leave "purely human art" as a very niche market share and it will be hard to compete outside of it if you reject it long enough. Also who says that there won't be an AI that will be able to fake "creative process steps" for every art piece you present it. Even before Dall-e/midjourney/SD boom there was already AI algorhythm that performed "neural style transfer" generating colored simple shapes on top of each other just like oil painter would do starting with big strokes and layering more and more small ones to add details.
@@cesar4729 I meant in the context of a hypothetical impersonation of a not so famous artist. Like an AI art thief account. But yeah that’s probably already a thing at the rate tech is going.
Start cryptographically signing and watermarking your images; pick a good enough algorithm, and the world would be in much bigger trouble if someone figures out how to break the cryptography than just having some perfect forgeries going around, so if it somehow happens, proving art authorship will be the least of your worries.
The thing is - if AI makes art-makers obsolete, the people who will be making money will still be the same sort of large companies who once employed artists. They'll still make money selling entertainment, etc... but it will be individual artists who can no longer be independent (and that's hard enough as it is). That's a legitimate concern... On the other hand, if it is the large corporations that become obsolete (say.. it becomes feasible for average Joe to make a beautiful 3D game on the level with AAA titles), then maybe it's a different story. But we're still more at a place where it seems like proprietary companies will benefit the most. Force 'em to go open source, take away their 'right' to use art not in the public domain, and things become much more humanistic and far less misanthropic.
@@sychuan3729 Who needs 'em? I could live without Fortnite and Mindcraft (kids would certainly benefit) and wouldn't mind a few more indie games and comics by small independent teams. The worth these 'evil big companies' offer comes at the expense of culture and value, as they skim the bottom for the most addictive, most gruesome (horror movies today...), most lurid, most vile.... yeah I have no value for these big companies or big Hollywood.
That's why people should support Stable Diffusion over Midjourney, or Dall-E. SD is open source and being open source has already has proven to be a superior strategy. Blender, for example, is already way, way more popular then the pair alternatives, mainly because it's entry barrier is next to zero when compared to the alternatives, but also because they added new cool things (like geometry nodes), but also allow for easy extendability. Overall always support open source over proprietary software, because this way the software is for everyone by everyone.
@@graythistle wasn't this one of the few points the two guys in the video agreed upon? An independent illustrator that has passion for painting grotesque realistic horror (what AI does best apparently) is not gonna suddenly have passion for composing comics or designing games for you to enjoy.
@@nocturne6320 I agree and think that the solution is Stable Diffusion's Open source model, which makes it a more reasonable proposition - the problem is that anyone can cut corners and replace an artist who has spent years cultivating a style that can now be ripped by AI. That is at best misanthropy. In all fields we should be protecting people from AI. This is what separates the Postmodernism hatred for humanity with a brighter future that will benefit all. As an artist, I am walking a fine line if I use the art of a living artist as a basis. I could easily copy a popular artist like Loish if I wanted to, but there would be a lot of backlash if I tried to make a living doing that - and rightly so. People would be telling me right and left that I'm a copycat. I walk no such line with the public domain and can mine it for whatever I want as I like - the AI and the person behind it should have to do the same, and suffer the same backlash I would if they don't. In fact, it should go further: that backlash should be legal and if AI produces a bunch of copycat Loish art, she should be enabled to sue the offending company into oblivion.
No one really mentions the physical material (severs) and energy costs (electricity) of running massive AI art farms. The more complex some thing is the greater the expenditure of resources is necessary to make it. So that silly 'mega feed' Steve mentions is kind of unfeasible, generating an entire movie off a conversation isn't going to happen it would require FAR to much power and render time.
Love how visual artist hope to be regulated like music artist. They just have to try what is considered "original" in the music industry, and then they will quickly change job because they will not possibly using any copyright material as reference or inspiration
The original video against AI felt like it was 50% hyperbolic wrong information and 50% condescending elitism mixed with fear mongering and alarmist propaganda. I stop taking people's arguments seriously when they start using massively manipulative emotional tactics, and most of what I could see being said against AI after that, is misinformed bollocks. I've had people steal my art before, both physically in person and online where they claimed mine as their own creation. I've had people sample my art for use in their own compositions and photoshopped collages. AI art isn't even on the same continent as any of those behaviours. AI really is no different than doing studies of someone's art until you can replicate their style, just as actual artists do.
I was already for AI art, though I was leaning towards the "should be opt-in" side of things. You made a really good point many times over about how we already opt in when we click "I agree" to the ToS. I have long accepted that for things like my personal data, but I never considered that extending to things like uploaded artwork.
Your uploaded artwork *is* personal data. That being said, it's only very small communities that actually try protecting their data, and I don't really expect that to change.
So, if some future AI technology scrapes the internet of everyone’s personal data and allows any random person on the internet to do what they like with it, you’ll be ok with it because you clicked ok to some TOS years before the technology even existed?
I think you're missing a lot here. This is a long comment, but please read it. 9:10 That comparison doesn't work, as LAION relies on images taken from the entire internet, and to my knowledge, the Google ToS doesn't include "Will be used to make an AI that can knock out images under your style". LAION takes most images from sites like pinterest and etsy. 11:30 There are only so many eyeballs for so much content. No one is going to care about someone's 300 pieces a day, that's not how attention works. The end result will be that FAR fewer people will be able to live off of doing creative media, with handful dominating everything. 1.4% living off of their art (will get to that number later, it ignores photographers, designers, and other soon automated fields)? Try 0.015%. That's the number of twitch streamers that earn as much as the average US household, or more. Given how streaming creates the most content hours for least effort, this is the best comparison, except that knocking out AI images and stories will be even quicker for more content. A future where basically NO ONE can live off of their passion. 14:30 The main gripe is how there is no protection for artists, like Greg Rutkowski. The "artists don't make that much of the dataset" argument only works if you ignore what the AIs to 95% are used for, and how artist's names are directly being used. 27:30 "Not using it is career suicide" -- So you're admitting that the vast majority of artists, designers, photographers etc will be put out of their jobs? It is easy for you to be so nonchalant about this, but don't wonder when Steven isn't sharing your ungrudging attitude. 28:50 His main agrument is about it taking images from people that never agreed to their usage in this way. You have yet to produce the ToS clause that allows this, and given how LAION is the base for these AIs, it's missing the point. 31:30 Of course they're not going to say that they're about replacing artists, though that is what it comes down to. The farmer analogy is disingenious and doesn't work. 31:45 "artistic people will be even better" -- I don't see it. It already pools from the greatest artistic minds, making far more decisions than are put in, and better ones than most could come up with. 35:00 He never makes this argument against automation, but I will make one. The more we have of something, the less it is valued. Why is it that during the age of instant access to great masterworks in beautiful colors, instant photography of gorgeous landscapes, we live in grey, uncared for homes, drive grey cars and use grey appliances? 41:10 So you envision a future where basically everything is done by program, and that's a good thing? Is it so weird for you to consider that there are countless people out there that love doing it themselves, and that have worked their entire life to get a job in this? When I see an art studio of hundreds of people that can support themselves doing something they love being replaced by one "quality assurance" guy occasionally checking in on the outputs of a machine that basically does everything itself, having more knowledge of current trends than any human could, that is simply dystopian to me. How far do you want to be removed from making something yourself? 1:00:00 "Feelings" -- Your side also hinges on feelings, because what is making stories or consuming them without feelings? I don't know why the idea, that some people love spending the time on drawing their own comics, putting in all their effort and care and doing every single stroke seems so alien. To me, the idea of letting a program do that is just shallow. If you really love your creation, wouldn't you want to make every single detail yourself? Your own argument is about how important the feelings of neverending consumption of AI content are. 1:04:20 Since diffusion models are prone to overfitting and memorization, you'll find bloodborne generations that look almost 1 to 1 like bloodborne cover art. 1:06:30 "Human theft is worse" -- that is simply not true. The most important clause of fair use is the 4th, about how cited work affects the value and market of the original. How is the value and market of Greg Rutkowski being affected by "cited work" being used to make unending copies of his artistic identity, to the point that he has stated that he worries about his employment? Greg is not going to care that you copied his image onto your desktop, he may even appreciate it. He may also appreciate a person making a work in his style as a tribute. Even if someone 1 to 1 used his work to get employed, he would barely be affected. That is "human theft", and it is completely benign in comparison to what AI "made by Greg Rutkowski" prompting by the millions does to the value of his work. 1:15:00 I highly doubt that they would've used copyrighted images, let alone made a program where you can enter an artist's name directly, if the artist industry was as capable of suing as the music industry. Do you really think the art equivalent of a Taylor Swift AI would EVER fly? 1:19:00 "Not a real artist" -- The crux of this argument for me is how minimal the human input is for (almost all) AI images. When you're actually drawing something, you inevitably have to make thousands of tiny decisions, that now AI makes. I would go as far as to argue that the idea is less important than the execution. Is a LEGO movie a good idea? Or a guy fighting crime in a batsuit? Not by itself, it's the thousands of small creative decisions that turn it into a worthwile endresult. I don't think AI prompts can't be art per se, but they almost never are. When you pool from the greatest artists that have ever existed, the AI is bound to make decisions that far exceed those of the average prompter in skill and uniqueness. Most prompting is only commissioned consumption rather than creation. 1:25:00 Please give me the ToS part where it states that images uploaded to pinterest or etsy will be used to make AI models that will make images themselves. Even for google cloud/photos, I don't think it exists, and again, LAION doesn't rely on google cloud, and even Google's imagen relies on Microsoft's COCO dataset and LAION. 1:30:00 "There is nothing stopping you from making art however way you want to make it" -- When I have to spend most of my life working a job that has nothing to do with art, for most of my life I'll be stopped from making art. You may say "That's 1,4%!", which excludes designers (fashion, covers, products etc), which excludes people with part-time jobs and photographers among many others, and which doesn't consider that the same thing is coming down the pipe for voice actors, musicians eventually too, writers, sculptors, actors, animators, and every other job that people actually love to do. 1:32:00 "If you don't make your own colors etc not a real artist" -- Refer to the earlier point about the majority of actual creative decisions being replaced. This comparison doesn't work. Don't be mistaken. This is NOT like any of the earlier cases. This is taking someone's work to put that person out of their job. If you don't find that inherently worse than coming up with a novel invention like the car, I don't think this debate will go anywhere. There is no historical equivalent to this.
the thing that makes me most optimistic is how unfullfilling prompting is. sure now it's a novelty, but what about a million generations down the line? I actually thought before this all came out at this level, taht it would be like a release valve with hundreds of millions of people instantly going for it, but i think most understand it's a hollow victory
Thank you for this comment, I read it all. I haven't watched the video. I did watch Zapata's last week or so, I liked his the best (I'm a hobby artist). It's not "fair use" if it takes a person's livelihood to train on, and then puts that person out of work. I have to wonder why Disney isn't putting their foot down, I saw a lot of Disney stuff...
Hey Royal Skies. I hope you respond to this, though I realize that you are a busy man. So I've been looking and talking to a lot of people in the AI forums (people on both sides of the fence) on reddit and stuff, trying to get a broader perspective on the subject. I've really been trying to put my mind at ease about all of this as the only real hope I hold onto in my life being able to find a profession creating art. As you are the one to introduce me to AI art, I was excided to what this video as I find myself having a lot of the same thoughts and feelings as the guy in the other video, and truth be told, after hearing your replies, I feel incredibly worse, and I think the big disconnect here is perspective. And please understand, I don't think you're perspective is wrong by any stretch of the imagination, but its missing a few angles, as is the other guy. I think where your argument falters is somewhat in combination of the feelings (and please hear me out) and gatekeeping. While I 100% believe that art and the ability to create should always be available to everyone ever, I feel that a gate is extremely important too and I'll use your examples to explain why. Going back to the Fire Fighter, the reason that they shouldn't be able to create great art so easily at such high levels is because they chose the path that doesn't/didn't allow it instead opting for another path with its own rewards. That Fire Fighter, like all Fire Fighters is an absolute badass! A Hero! And we/society all know that and acknowledge them for it. They can lay their head down at night knowing they made a difference. They can do that because they chose to spend their time becoming great at that skill set. Artists can't run into burning buildings again and again and be those heroes because we didn't train for that. Its a gate we didn't pass through, and that gate keeps unexperienced people safe from the flames. Artists chose to spend time acquiring art skills instead, and whether people want to admit it or not, I'd imagine all artists want as many people as possible to see their work and praise them for it. It makes us special. The Fire Fighter can raise their head high for what they do, and we can raise ours for what we do. And I think this goes especially for artists as the majority, at least from what I've seen in art school, forums, ect, grew up as the "not cool" kids in school. This is supposed to be what makes us special, what helps us shine, but AI Art is like a super power and in the words of Syndrome from the Incredibles, "Everyone can be super! And when everyone's super, no one will be." The other major part I feel like you really, really glossed over is the fun factor of art, but I think I get it. You are an end product kinda guy. That's cool. Nothing wrong with that. You're a badass, a successful badass (so you're lightyears ahead of me). You be you, but you aren't seaming to realize that not all of us are. You ended the video saying that artists will always be here. That we'll always see that kid doodling in class, and you are 100% right, but that's because grade school is boring and drawing isn't. The dream (and I do understand how much work being in the 1.5% of profession creators is, I'm not low balling it here) is to get to be paid to do that. To walk into your studio and just create for hours on end, to be paid to basically have fun (you know, that saying about if you love what you do you'll never work a day in your life) and for some of us, creating is the process as much as it is the end result. I'd love to just get paid to sit and model a dragon for hours on end, but if the AI does the fun stuff for us....how is it any different than any other job? Hell, you even said that prompt crafters will probably just end up as AI managers down the line, and maybe I'm alone on this, but manager style duties is almost the opposite of being an artist. One of the big selling points of artists isn't just that they make great art, but that they could make it fast. It gave them leverage for pay. That's gone now. So what's going to stop companies from cutting artists down to a skeleton crew and lowering pay? Once those doodling kids in class see that drawing can't make them a living, well they're probably just gonna chose another field, especially if they can just toy around with AI. I dunno. Maybe I'm way off base here, that's just how I personally feel right now and really needed to get it off my chest, and while there's more I could say, it'd be mostly personal feelings and short sightedness. Thanks to anyone who read this far with an open mind. As a lousy artist who can really only benefit from the AI (and I do see all the cool ways to use it), I really, reaaaaaly hope AI turns out to be nothing but greatness.
This. I agree with everything you said here. strangely enough I was reminded of that exact quote from the Incredibles. You can tell when someone put their blood, sweat, and tears into something they are passionate about. I hate the fact that i spent so much time learning how to model, texture, wight paint, UV, and several other skills just for an AI come in and do it all in a matter of seconds. Its extremely frustrating and makes me wonder if there is really even a reason to continue doing what I love. Maybe I'm being dramatic, who knows.
@@sterlingbirks9101 Yeah, when those college bills come in and there's no job in sight, it can be rough. I'm still paying off my loans 10 years later, though I made lots of mistakes so I'm not a good example. Still, I feel you. Royal Skies is super cool and super talented, but doesn't seem to understand how this threatens people who aren't as talented as him. If it helps, I've see work flows that use AI art in ways that still require effort and skill so I'd say look those up so you can still get some satisfaction from your efforts.
If you enjoy the act of making art, nothing will stop you from continuing, in the same way that some people still make their own clothes, or blow their own glass, or whatever. And some will pay a LOT for that, despite automated techniques that exist to produce both.
@@sterlingbirks9101 The worry is mostly unfounded. "Specialized real world art" is ALREADY the only way artists make a living, regardless of genre (visual, audio, performance, etc). No one, except Redbubble cell phone case folks etc, has a huge market for "pretty pictures." They need something very specific, to fit into a workflow of often dozens of others, and the AI is no where close to closing that huge loop on its own. Source: I've made 100% of my wage doing art professionally, for 25 years. People who focus on craft over concept rarely go far, because no one wants boring art that doesn't hit the brief, even if it is well-executed.
I know this video is trying to keep the ethics confined to AI art in a vacuum, but the "Consent to EULA" argument is an appeal to Status Quo argument. The reason people skip past TOS agreements is because all those agreements contain more or less the same stipulations, and the only alternative to agreeing is to not participate in modern society. This also runs counter to the ethical underpinnings of intellectual property, which was initially conceived to stop people from just stealing newly created works.
Yeah that argument was a bit weird to me as well. Many of the websites you upload your art to don't have anything to do with those companies making these programs. So I doubt those terms of services include being ok with ones art being included in those training data sets. Who knows how that's gonna play out though and if using someone else's pictures is considered "stealing"
@@discipleofschaub4792 Like, would he make the same argument to Eve6? A 90s band that sold the distribution rights to their songs before Spotify made those rights worth more than the CD royalties?
"But if you didn't want your work to be seen by anyone and anything you shouldn't have make it public." Thoughts like this is why AI ethics is a thing.
@@maxicornejo9675 AI ethics is stupid and just a way to preach certain ethical positions when they only signed up for the knowledge part. I'd make the case that if I had read Harry Potter, that likely, in some way, affected my ability to write english, and that, in some way also affected my writing style. In essence, ai isn't stealing any more than I am.
@@Cecilia-ky3uw I have three questions for you. Do you think merit is important? Do you think that an artwork provides more value to an algorithm than to a human? If you do, where do you think that value comes from?
I think the counterpoint from you against stolen art is pretty much a straw man. He never said anything about hacking or uninformed platform usage. I agree with you that people give out consent how their own work can be used way too easily (let's say they post it on Instagram for example). But the issue is still valid. Those neural networks are trained on art without consent. You can actually look up the database used for stable diffusion because their set of tags and links is open to use. However when you look up the actual images behind those links, they are not owned by the developers or researchers. Those are public images in the internet (public as in accessible) but there's no information about the license to be found. That means legally in quite a few countries, you should expect them to be copyrighted and therefore they should not be used for training of AI without explicit permission. Otherwise every artist on the internet would need to build up private paywalls and start suing other people like crazy to counter that. It's a massive copyright violation and it needs to be addressed. I'm very sure of this because Github copilot from Microsoft did the same thing. They just trained their neural network on public source code and sell the tool now. However it is impossible for them to always check licensing for each piece of code. So it is extremely likely they violate licensing by selling generated source code derived from code which potentially explicitly denies that by license. Then if a company like Microsoft doesn't care about this, why would any other company request permissions for artworks here? You don't need to hack anything. If you just crawl all the images from DeviantArt to train a neural network, you still violate copyright. It's that simple... would be a totally different story if they had only used images from Pixarbay or similar.
BTW in copilot's about page it was mentioned that the system "shouldn't" generate copies of existing code or public personal data from training material.... "Copilot was trained on publicly available code..." and some dancing around the issue of ownership... "GitHub does not own the suggestions GitHub Copilot generates. The code you write with GitHub Copilot's help belongs to you, and you are responsible for it."
No it's not. Training using existing art does not violate copyright. The same how sampling a popular song and modifying it doesn't violate copyright in the music industry. There have been court cases regarding sampling, and more often than not it gets overruled as a moot point. The AI may use your art to train, but it is not giving out the art to others with a copy and paste, nor is it using 'commercially' because the output is always modified. There is no copyright over art styles, which is also a popular point. If someone makes an AI piece of art that looks and feels like your art, its till not yours.
@@devnull_ Of course it doesn't generate existing copies. Still it generates portions being copied. There has been occurrences where authentication tokens were suggested by copilot which shouldn't have been publicly on Github anyway. But that's not the point. AI algorithms are pretty much always trained without consent of actual owners. Because it wouldn't be feasible to do so. The training sets are huge and asking every owner of data for consent would take years. So this is either done by terms of service or illegally. To me it's a whole reason not to use Github if possible because I work on free and open-source software on the assumption the software stays free to use. However if copilot gets trained by such code, Microsoft gets paid for it and the result ends up in proprietary software without publishing its code (even though portions might belong to me and are licensed under GPL)... well then AI is effectively just a tool to make license violations legal, isn't it? So from my point of view none of the algorithms are problematic. It's totally fine to use AI for automatic generation of code, images... whatever. But the training should not hurt the original owners of the data. However currently it does. I personally would love to see open software which allows artists to train a network with their own images to help them reduce time doing their job. But that's not what is marketed here...
@@DankaDoctor185 It's not safe by copyright either, though. The part that people seem to always miss in these conversations is the word "USE". It's not just about the image itself, it's about the usage of the image. Sure, the AI might not carry around your photo or drawing verbatim, but if it knows enough about you to recognize your name and make a doodle that looks like something you did, that's usage of your work. Sampling and modification are also only fair use under HUMAN rulings, you cannot apply the same logic to an algorithm, there's no heavy legal precendent. Lastly, since your case is about fair use, I feel the need to let you know the following is a factor in court decisions of fair use: "The effect of the use upon the potential market for or value of the copyrighted work". Now understanding what use means in the legal sense, suddenly this is not as clear as you make it out to be. I'm not going to claim one way or another if it's truly legal, but I will say this point would probably not hold any water.
@@TheJackiMonster Because I like keeping things neutral, I do think it's important to state that the AI does not actually jigsaw existing images together. It kind of works in a way similar to how your brain constructs things. th-cam.com/video/1CIpzeNxIhU/w-d-xo.html This video is a really good explanation, but the argument that it's actually using other images and reassembling them in portions is incorrect. It's a really fascinating technology, and while I'm against its commercialization at the current state, a model trained on Pixabay or another CC0 database could be incredible if the steps were taken to filter out copyrighted material.
I have to continue watching your video tomorrow. It is far to great to watch it at 4:30 am on new years day. But I wanted to write something about myself and why I am glad about AI being as great a tool as it has become until now. For me, AI-Art-Systems are plain hope. I am severely disabled due to a multiple sclerosis, have two partially paralyzed hands, cramps and spasms and worst of all: fatique. I had to give up my dream job as an art director in a small advertising company - in fact, I am unable to ever work again. I had to retire and can‘t really get out of my bed and even less my home, without help. I still make art, it takes a huge ass time to make art, but I need to express myself creativly. i wouldn’t have any wish to stay alive with all the chronical pain and being unable to live on my own, if I couldn‘t make art. The problem is, that I won‘t be able to use my hands indefinitely. In the moment, I can skip a lot of time, when using stable diffusion as a tool. When working on the right wording to get exactly the picture, I would paint or create myself and then complete it with my own paintings. In the last few month I more and more developed my own workflow with it and it is amazing. If I would still work as an art director, these systems would make my work life so much easier, giving me time to create even better works for my clients, although I also think, that many companies will employ less artists and focus more on the directing creative people, rather than the graphical artists themselves. There will be new jobs and new opportunities for artists in the future, but yes - some professional artists in the industry will loose their jobs due to better and faster solutions. That is sad, but it happened in every branche, with every job so far over the time. Artists used to be the lucky few, who would never be replaced by a computer, because they cannot create art… oh well, that is now a thing of the past - but this doesn‘t make our challenges special, it finally makes artists like everyone else out there. To come to a conclusion for today: Steven sounds like he speaks for every artist in the world but only speaks for himself and those not understanding the matter. Either out of fear or with vicious calculation, he shows a total lack of understanding with a pretentious and gatekeeping attitude. And he (in my opinion) reached his goal. Finally everybody talks about him. Sadly nobody about his great drawing in the video and everybody about his lies in one form or the other. oh well…
I generally agree with you point view, however it seems like there will be an escalation of a big purely digital problem which already exists now - informational entropy. Even now it’s extremely hard work to find a useful information for self education, you have to scroll and sort a tons of bits in order to find a truly quality inside shared by someone. And with all surrounding ai-technologies the amount of produced information will grow exponentially faster than now. There will be no place to search, just pick the best solution proposed by the algorithms trained on your data. I think this is a huge problem, not in a personal level (because personally it’s amazing to perceive a purely individual things you love), but it will apply to the commercial industry of all sorts and cultural mess. Maybe chip-implants is a solution, but not a certain one. I hope people will find the ways to spend more time off-line with each other, but for now, it seems like atomization to reality is an up-going trend.
Good point. Which I'm sure one of the big software companies are working on already. Ai art already is trained on good vs bad quality so I'm sure we aren't far from the idea of search engines filtering out the most accurate data we are looking for from trillions of data.
I already notice that (such as in online shopping). I think there will be a number of solutions including going more personal and private vs. just cruising the internet and social media.
The first thing I don't like about AI that you don't have much fine control about the creation as it is now. That's why Zapata is right when he says that it isn't a tool. It is only a tool in a very broad sense, but it doesn't stimulate us creatively and mentally the same as when we are way more involved in the process and in control of the microsteps. So it hinders self expression that way and spreads mental laziness. My second problem is that AI will lead to a massive oversaturation of the art market (and possibly movies, music, video games etc in the further future) so it will become increasingly difficult to stand out. You are right in that the truly great artists could benefit from that and stand out even more but that depends on how good the AI will become. Should it reach a certain level of perfection or would even surpass human creation then together with the oversaturation of art everywhere it could lessen the general demand of and interest in art to such a degree that the spotlight artists often relished until now could dim or go out completely and I don't find that motivating for artists, who want to get attention and popularity. You say that art matters even when you are the only person in the world (maybe together with a few friends) who likes it. And while that's true, it seems a bit involuntarily cynical to me because it's also extrinsic motivation that drives everything we do and we also need external validation to be complete in our happiness. And the prospect of a world where all your artistic efforts are in no real demand anymore is just disheartening and hopeless.
@@dibbidydoo4318 It's the same reason why having a (good) job gives meaning in life and having none often leads to depression: Because you make or do something that is of value not only to you but also to others. Humans are social beings after all that don't live only for themselves.
@@dibbidydoo4318 You and Royal Skies does not realize it, but just because you don't think it makes you happy does not mean it does not make you happy. When you make something, put a lot of work in it, you are probably doing it to share it with others, this goes beyond painting and such, even just the idea itself. If all the people you will be able to show your idea are just your family, then you will not want to spend any time on those things. You will have creativity that you cannot release because back in your mind it will not matter. A child seeking validation from their parents is normal, but not when you are older. People need validation from others, not just in art, but in most parts of their life, but especially when creating something.
@@lexiferenczy9695 I agree. There seems to be an idea going around that working to make other people happy is a kind of narccissistic attention whoring, but working only to make yourself happy is enlightened selflessness. It seems obvious to me that these beliefs are cognitive dissonance.
Could you point me to the section of the Icloud user agreement that gives third parties the right to trawl private cloud data? I can't seem to find it... This isnt how the AI data sets is created by the way but I'm fascinated by the multiple directions you're able to wrong at the same time.
Agreed. No idea how he keeps going on and on about "you agreed to the ToS" when there's absolutely nothing in the ToS that would permit it, and as an "AI artist" he really should know better about where the data comes from
Man if only the world was at a point where it can openly embrace this without having to worry about having a home and food, because that's what it boils down to in the end isn't it? You put years into learning your craft so that you can eventually live off of it, all the while AI slowly but surely is automating jobs away and changing fields isn't simple or easy, and we still don't have housing for everyone or food, plus it still is pretty unclear, uncertain, where we are headed with automation. Whether we'll actually be free to do as we want or everyone will be cast into poverty as they're no longer needed.
100%. Automation wouldnt be so terrifying if it wasn't eroding people's jobs, and arguably at a faster pace than new industries are rising up to replace them. We don't live in a post-scarcity society. The resources needed to feed and house and clothe and care for everyone on the planet exist, but not everyone has access to them, and they are making what you need to live harder and harder to access every year inflation+cost-of-housing+cost-of-food+cost-of-utilities(because inflation includes none of those things) grows faster than wages (which is every year in aggregate, and most years for most of the population)
This is why humanity is going to HAVE to get past this "you only deserve to live if you're slaving away 40 hours a week" thing, and implement something like a UBI. We're gonna need to stop demonizing "lazy people" because EVERYBODY is going to become obsolete, and it will be happening faster and faster. People in the US rant and rave about how much they don't want communism or socialism, but that is exactly the kind of things that automating everything is going to force us to embrace!
You know what is funny about that argument? Is when you look at history. People complained when machines replaced factory workers. However, before machines replaced factory workers most people were never given the chance to aspire to be anything more than that level of worker. Only the very wealthy could afford to be artists and inventors. Poor people stayed poor, they weren't offered a very good education because why would a factory worker need a good education? However, at the time people didn't know that machines taking their soul destroying jobs would benefit them and us, future generations. Imagine if they had really fought against it to keep people in these kinds of jobs. Our generation would still be there, and only the wealthy would have the luxury to be creatives. My great Grandfather was a coal miner, so were all his children, apart from my grandad who was born as they banned child labour and instead children of all classes were given an education. People complained about banning child labour too and sending them to school. Before people had children so they could send them out to work, children were made to start working at a very young age. It was only possible to offer all children an education because machinery were starting to be invented, that mean they didn't need everyone working. If anything these programs are being used by people that can't afford lots of art materials and equiptment, its not going to prevent anyone from creating art if they want to and its not going to stop people buying art made by people, if they like it. All it is going to do is give more people access to creating art. Exploring you imagination shouldn't be limited to only those that can afford it. I don't know what you mean by no longer needed. Art is always trending so what if your art style isn't currently in fashion. That shouldn't stop you from being a creative. Humans have been creating art since cavemen painted pictures on walls, it is part of who we are as a species. Its not going to stop just because we have more tools to be creative with. Even though most people now have access to cameras on their mobile, it hasn't prevented professional photographers from getting work. I know a lot of professional photographers that once worried the smart phones would take their jobs. It hasn't. But it has allowed everyone to explore photography and discover if they have a talent or passion for it without needing to invest in expensive equipment. When the camera was invented lots of painters freaked out that they would lose their jobs, thinking why would anyone hire a painter if they could just take a photograph. Yet painters are still around, if anything there are far more painters now than ever before in history.
@@codeXenigma I get what you mean and the automation of past industries is a fair point, I frankly do not know what to expect so I'm just wait and see, and learning a bit about this AI since Royal may very well be right in not using it being "career suicide". As for no longer needed, I mean that this may cut down several jobs (I mainly see talks of concept artists in regards to this) but of course not only in this field but many, yes it is true that as one thing gets automated and those jobs are lost, other have come up, but it remains to be seen how AI will play out, I do fear Steven may be right that this is not a tool but a replacement, sure you can keep doing it as a hobby and AI does open more creative opportunities to everyone, but if you can no longer sustain yourself because your field has been automated, you are indeed no longer needed, so change fields, and hope that also isn't automated away? And hopefully still have time to be creative.
@@codeXenigma you lose me as soon as you compare companies ripping off artists all across the internet, to child labor... Art is one of the very few jobs that people WANT to do themselves. You're really gonna compare 30 years of passionate training for something you love, to being forced to work in coalmines?!? Apples to oranges dude. You think ai art is no different than when calculators came out? Well calculators didn't enable people to go around pretending to be Einstein himself. There is a WORLD of difference.
1:25:30 I want to be charitable and say that I think you misunderstood the entire point here. If I say "I didn't agree to have my work put into this AI databse, it's sus that they never asked me to agree to putting my work in the database, and I know they will not ask nor allow me to opt out in the future, therefore this bad and doesn't care about me as an artist nor my consent" you responding with "just be happy that they gave you this opportunity in the first place," is equally immoral if I take you at your word and that is in fact how you view this. If someone doesn't want their work utilized in an AI database, that is their decision and it should be respected, whether it "grows their audience" or not.
Do you have to explicitly consent for another artist to view your work that is publicly displayed and then use it as a reference to produce a completely different piece of art though? No, you don't. The consent is already implied.
Really appreciated this upload.. its right on the money. The Laion training model being used is an unethical montrosity designed to smuggle 5 billion pieces of human made art in such a way as to specifically make litigation very difficult. The end result is nevertheless no less damaging to peoples livelihoods. To quote S.Gordon's film, From Beyond.. " "that machine has got to be destroyed"
I see art moving in the direction of youtube, etsy, patreon, twitch etc. More artists will have to become internet personalities/commission artists. And work with ever day people directly instead of working for large companies or studios. I don't think the mid or higher level commission/entertainment artists will be very effected by this. Commissioners often buy the art they do because of the brand or person behind it. For example, look at closed species communities. People pay just for the "right" to draw a closed species character themselves(not even art! just the right to draw it themselves), and sometimes hundreds of dollars for premade designs just because of the brand behind it. It makes them part of something. Its a badge of honor, a bragging right. It makes people happy. AI art can't hold that level of prestige. If anything prompt engineers are going to receive the same kind of backlash digital artists used to get... Constantly being told their work is "not real art." And having its potential value be scoffed off by the mainstream. I do think beginner commission artists will be effected tho. I used to sell my work for next to nothing ($5-$30 range) and the clients I got back then where often a LOT more unpleasant than the ones I have now. (some even trying to scam me into working for free). Looking back, they obviously bought for the bargain , not for my style or quality. I can see those types of people using AI for free work instead. (Gone may be the days where artists are asked constantly for "free work"). But that does leave a smaller client base for newer artists. They may have to work longer and harder before they can sell their work at a reasonable price. I don't think people will suddenly stop supporting artists and all swap to AI. If anything it just gets more people into art and makes it more accessible to more people. I wont sugar coat it tho, a lot of artists are going to have to change the way they do things. The quality expectations are going to rise and Ai brings a whole new level of potential thievery/con artistry into the mix. I do find the click bait articles exhausting, and the copyright issues give me a headache. The companies leading the way make me nervous as well. But I don't think we should all just give up or spend hours on end arguing over "what is art." (not that anyone here is doing that, I just see it A LOT). If anything we should be challenging the people at the top, and making sure AI stays accessible to EVERYONE. And not just big corporations. The goal of AI is to make things easier and more acceptable and I think we should be working hard to keep it moving in that direction.
We need to fight the corporations on this one. If it's ruled that "style" or general output can only be Owned by whoever owns the rights to the software or the rights to the training media, all they'll need to do is what they've already been doing: either buy out the AI company and put something in the EULA that ensures they have a stake in what's generated-- or otherwise have right to limit the output; or pay an artist to produce the art they can train the AI with as long as the artist signs away their rights to what they produce under the company banner-- and they'd simply outsource the creation to talented artists in countries where the cost of living is lower, as the corporations usually do, so they're not necessarily beholden to paying a living wage. Once AI is accessible to everyone, corporations can't hold culture and entertainment hostage as they already do with immoral copyright laws. The people for whom it's important to see "people like them" on screen can make it happen instead of paying a "Creativity Landlord" for the privilege of being seen.
Thing is what are those people going to pay with when they are replaced too? And why put another strain on the wallet of those who need it the most instead of the billionare companies already siphoning all the wealth out of "the people"
Just so you know stable diffusion on automatic 1111 is 100% free. Open ai created it exactly for the reason to give access to everyone. Also the client base may shrink but your work also becomes easier if you are using ai tools.
Undercutting new artists from ever trying to get good. Early artists tend to need some hint at financial success and the existence of image generators will probably create a shortage of competent artists in a probably three generations.
@m f I don't see anything wrong with that. Markets change. And those brave souls left will become invaluable. Everyone else will still draw and paint for the joy of doing it not for profit. Jobs don't have to be your passion. It's sad yes but times change all the time.
I don't even think AI art is as great as some people seem to think it is. At least when it comes to the art I care most about, most of the time when you look too closely, you realize it's an eldritch abomination of some kind and during the few times when it's at least passable (usually when the places where it has the most trouble are hidden somehow), it's always really simple and in the same few styles with basic perspective and no action or idea. Besides that, AI art isn't really capable of most of what people who commission art really need. AI can stay, there's no reason to argue against it because it's already here and isn't going anywhere. But I also can't see it being used for very much.
it will get better, the hands and the slightly uncanny eyes and the illegible text. Unless the tech hits a significant plateau for some reason in the very near future (unlikely considering renewed massive interest and funding), these aspects will all be improved upon quickly
@@andersonrobotics5608 Can we really be sure it'll happen quickly? AI can easily hit a significant plateau. For example, people are excited for language AI, too (things like AI dungeon, dreamily, NovelAI, CharacterAI, etc) but it has been struggling with the same shortcomings for years now, or at least, the applications of it that are available for me to try have. Some of them have made small improvements, while still conceptually being flawed in the same way, which is that each of them have no true memory and rely on a big wall of text for context, instead. This wall of text becomes larger and larger over time, which saturates the context and confuses the AI, before meeting a maximum character/token limit and being truncated, causing the AI to forget important details while also making up new ones that usually don't make actual sense. When it comes to image AI, the problems we're seeing now have always been there, because AI doesn't know how to compartmentalize its data to specific parts of the human body and refer to them while drawing. The AI also can't tell when it's made a mistake and go back and fix it. But even then, the human body is only the tip of the iceberg when it comes to all of the things the AI has no idea how to draw, and especially in different artstyles. It will try anyway if you let it, though, so we end up with people carrying living, soft guns with multiple magazines and weird gun-like shapes sticking out of it, or cars walking along the road with legs. Which is pretty funny or disturbing depending on how you look at it, but I have no idea how it would fix these issues since there are an infinite number of different objects that can appear in a picture in just as many different perspectives or shapes or kinds. This problem, as well as hands (or feet or faces, etc), could easily become the plateau, similar to language AI.
@@andersonrobotics5608 So far, that is just an assumption, and we cannot make a judgement based on assumptions. Will AI get better? Definitely. But how much better? Even at the level it is today (which is underselling how impressive it is), it has very obvious flaws.
@@ShiaNeko you say that the AI doesn't know if it made a mistake or mess but that doesn't mean it won't be used very much, simply that it will go through multiple iterations and humans will be judging and tell the AI where it made a mistake. I think text generators or language models is fundamentally different from image generators so the growth of the technology won't be similar. This is the difference between Qualitative progress vs Quantitative Progress; quantitative progress is something you can graph on a plot like computing power but qualitative progress like image generators and language models is impossible to plot on a graph because we don't know the limits of the technology and if the problems are against the laws of physics or simply learning a technique to solve it. For language generators we are improving but just because image generators and language models are run by AI doesn't mean the problems are the same. Just like batteries/energy and computer technology did not progress the same rate.
“AI makes art generation too simple and allows even ‘unskilled’ people to make images”. I am sure the same was said about the first cameras and later digital cameras. Does that mean you would like to argue that digital photography is NOT an art form? If you agree that digital photography is an art form, does that mean that every selfie, insurance estimate pic, and accidental pocket snap would be considered art?
I think the moment you start arguing for automating art, you should realize that it's very anti-capitalistic to do so. The reason is simple: The automation will improve and there will be the point, you won't be able to make any money from your art. Because it will replace your craftsmanship the same way factories have replaced many workers in other jobs. The reason I personally find going this route disturbing is that artist already have a very tough job to live from their work. Because art is subjective and art is not really something people can order. The AI messes with both of that things at once because now anyone can order whatever they want. Today an algorithm can paint an image, tomorrow it can make TH-cam videos or TV shows. Technology usually doesn't improve in the speed you prefer. So when you are going this route, expect it to go nuts in every possible way. Because you might think, you are not just a creator of art but also something like a teacher on a video platform. Well, I don't see why AI would do art in the future but stop right before educational content. Then of course we can think of the benefits. Everyone could get access to that, right? Sounds good but I would expect that any algorithm will be built behind a paywall from a company. So pushing those algorithms will likely result in monopoles of art as well as other content and people doing anything comparable the manual way will suffer from that because they can't keep up with AI. Honestly I personally don't think that it will bother me as much because I learned computer science. I know how to write such algorithms automating things for me or others. I expect to be able to keep up with this. But anyone who thinks that those algorithms will simply stay tools the current artist can use to improve their work is delusional. It would be one thing if the AI was trained on pieces from artists who actually sold the right to those companies using the AI. But from my research that's not the case. Of course all of this wouldn't be an issue if we wouldn't live in a world where everyone needs to make money to survive. But that's how it is at the moment and I can only recommend artists to start looking for a different field of study. Because why should people come to you auto-generating images when they can do it as well? Sets you up in a very bad spot for negotiation, doesn't it? Bringing AI to the table is not a new tool, it's a competitor who can steal experience from others in any form or scale. Good luck competing with that.
Half way through this. I love your videos and I have watched every single one of them. But I think you need to watch more of Steven's videos, he is usually never this critical in fact... never, you can tell it comes from intense frustration of watching things unfold hurting a magical world he loves. I have many friends and students I was mentoring who were studying art and suddenly gave up learning to draw when they learned an AI can do it. A depressing amount. I am somewhere in the middle I love AI and use it in my day job, but I do fear for young artists that feel like they need to use an AI to be competitive and may use it as a crutch. Like an aim bot in a fps, it would loose all fun. Every artist wants AI to take tedious tasks like rigging and weight painting. But the act of painting and designing stroke by stroke is a blast! Be like telling dancers that are lucky enough to be paid for dancing after working for years on it, that their dancing can no longer pay their bills unless they become a choreographer.
I think eventually people will be able to rekindle their joy of creating art, and be able to differentiate the experience of creating art compared to a robot creating art. An example would be there are still many people whp enjoy making jewelry, even though a machine can create the same ring or necklace. I think society will naturally value "hand crafted" art over AI generated art so maybe we don't have much to worry.
Props for the hard effort put into this well commented and personal presentation of the comments made by the creator of the other video. I didn't find a single point where you told people how they should 'feel' or tried to leverage the discussion. As someone who's spent years trying to make art of things I conceived and failing for a multitude of reasons, I find Stable Diffusion a complete life changing tool for my own gratification. Let me be completely clear that in the past I have commission artists, thousands of dollars over the decades to render in some fashion my concepts, most did amazingly well, a few never completed the work and ghosted me, and many needed to redo their work because specified components of my request were not present, poorly rendered or just not in the spirit of the piece. Most complied but some told me that's as good as it gets for what I paid, "they have to make a living after all." Then there's the ideas too far afield to ask someone to create. Stable Diffusion has managed these with surprising results that in some cases where better than the original concept or pleased me well enough I felt the task was completed and that was that. Would I use an AI picture to help provide a reference to an artist to have them craft it to my liking? Absolutely, it would give them a clearer idea of what I sought and even provide details, words are want to describe or omitted. At no point in our current AI technology, and quite likely far into the future; if ever, would an AI be able to write this kind of a comment, I feel safe to say, where the presenter in the other video feels sure now AI could render as good or better. There will always be aspects of art, in whatever form it takes where people will be able to do better than an AI. Tools like Stable Diffusion are tools, people create tools to make a job easier, allow things to be done that an individual would not have time, skill or drive to do, etc. No one ever said, "Don't use a lever to move that stone, hire my brothers to move it for you, they need the money." Tools generally are created using pieces of ideas other people had. Over the span of 5 weeks, I've used Stable Diffusion to make about 200k images, out of all those images I've had 7 I thought qualified as meeting my sense of what was artistic and suited my concept. Imagine if I'd paid a fleet of artists for that many images, how cheaply would they have to work for each image to fit my budget, how would anyone else find a artist to make their dream become a vision?
Already said this to Shad, but thank you for breaking the VERY one-sided narrative against AI art tools, Royal Skies. This should be a discussion, but one side has turned it into a moral imperative dogma that can't be questioned, unless you're a "bad" person. The dogmatic echo chambers/panic mobs online get on my fucking nerves.
Thanks for this video. It is nice to see people that can think logically and clearly. And accepting that others have a different opinion than you do, and that is fine, is rare.
Personally, I can really only think of two issues with AI art. I myself use stable(waifu :)) diffusion for inspiration in my own pieces and it works great. My first problem is that AI art will incentivize people to stop learning about art. Instead of taking the time and effort to learn about things like color, composition, lines, etc, people will just type "masterpiece" into a prompt and get a perfectly good image. I don't see the generation by itself as a bad thing, I just don't want the knowledge of art making to be phased out from society. As an anecdote, I want to make an animation with original music. As an animator, I already have the knowledge to do the first half of that equation. What I don't know is how to create music. I want to learn how to create music, but after learning about the development of dance diffusion, I no longer have an incentive to go out and actually learn. I don't want this to be the future, where nobody has any clue about the fundamentals of art(including music and everything else I might be forgetting) and instead just type a prompt into what is basically a search engine. In this video, you talk about how you want to spread your knowledge to everyone, but what happens when that knowledge is no longer relevant? I can't see any solutions to this, or if this will either be a problem, so people much smarter than me will probably have to discuss this. My second issue with AI art is the theft of art style. Just recently, after the death of legendary artist Kim Jung Gi, someone made an AI model that can perfectly copy his artstyle. This left a bad taste in my mouth, as well as many others, as an artist's style is basically their identity, and the AI has turned that into an algorithm. I have no problem with using AI and attempting to recreate the style of an artist you admire, as that's just what humans already do. I have a problem when you take hundreds of an artist's works and load them into a machine to pump out thousands of cheap imitations of that artist's work, especially if you claim them as your own. As to why I don't mind hand copying or even using an AI to copy their artstyle, you still need to input your own ideas and are using their style as inspiration. I don't think I need to explain how describing Kim Jung Gi's style perfectly to an AI and using aspects of his style to improve your generations is much different from training an AI to specifically copy that artist. If these two issues are addressed, I would be much more comfortable with the upcoming AI revolution.
@OliAnims These are interesting points because the discovery of computers has opened a way to more sophisticated information processing that does lead to style, form, and expression, the kinds of things we identify ourselves as. But now, the question becomes much darker: If style can be accurately reproduced, does that imply it is merely data in our DNA that enables each person to present a style, or is style something more 'personal', whatever that means? Also, if style can be 'observed', does the Universe have a style which we can characterize? Or does style exist only in the information itself, and that is what the AI is extracting from the images (that is, patterns), and not seeing style in the image itself? The second issue is whether the AI can replicate the person's personae well enough to create a history that isn't real, such as crimes and other nefarious actions, basically identity theft? It may be the impetus for us to re-examine ourselves and our purpose as humans, life, and biology, rather than reject the technology. Interestingly, we have turned full circle. The very copyright laws incentivizing people to innovate, learn, and progress are being bypassed and replaced with knowledge already gained through copyright. So, do we have sufficient knowledge for machines to do that for us, and we instead repurpose ourselves to more important activity? Almost every program written is imparting human knowledge into a machine so others do not have to know how to do the tasks. The AI is no different except that it is addressing the knowledge space itself, drawing patterns and inferences we, as yet, cannot.
@@noreply3322 "AI draws for you." What do you mean by draw? The AI is a program that samples images and reorders the samples into shapes that is applied to an output image. That isn't the traditional definition of 'draw'. The program links text words to visual shapes and those are linked to areas in images that the AI can sample and 'reproduce'. The AI composes images after 'learning' how millions of images have been composed. There is noting inspirational or passionate in that, nor creative. And by definition, Art is inspirational, passionate, and creative. The result is amazing and often stunning, but the AI isn't drawing for you, imho. If your friend is the one that drew the cat, and the AI copied the drawing and learned how to 'draw' a cat from it, is it your friend's cat that the AI 'draws', or its own cat?
@@petergostelow I personally believe that an artist's style is a culmination of everything they've experienced in life. It's not just a mashup of all the art they admire, but also of their personality, their childhood, even something as simple as what they are for breakfast can have an impact on what they draw and how they draw it. AI, and those who use AI to copy, don't take any of that into consideration. They take the surface level patterns that an artist likes to use and make a mockery of everything the artist is. Now, I understand that the person operating the AI has THEIR own experiences in life, but none of that is shown in an AI generated piece, especially if it was SPECIFICALLY trained on one singular artist. Imagine having to not only helplessly watch, but also compete with a bastardization of all your life's work.
@@petergostelow DNA has nothing to do with "artistic style", DNA is a hereditaty data storage, mostly responsible for storing protein structure data. Artistic style is not hereditary trait, so it has nothing to do with DNA. Stop using scientific language you barely understand. Artistic style is a combination of aesthetic preference and art training. Artist trains to use instruments to copy and modify existing art to their preference, after they master the instruments and art techniques it's just a matter of preference.
@@xn4pl "DNA is a hereditaty data storage" DNA stores a replicating program. It is responsible for building everything, including your brain and how it is wired. Therefore, personality and characteristics are also embedded in the code. If not from DNA where does the personality and preferences come from when you are one cell in the womb? Do you know? Each person receives about 1/2 their DNA from their mother, and the other 1/2 from their father. Each child has different parts from each parent so they appear different, but the same. The parents characteristics are passed on through the DNA, modified and adjusted, so the 'family' share common personalities, art style, and preferences, but none exactly the same, not even for identical twins. Where does personality and art style come from, if not the brain, and where does the brain come from, if not the DNA? If art style were not a pattern and procedure, it could not be replicated in a computer program, could it? Well it is being replicated as numbers in AI. Both the AI and the DNA are numbers arranged in a specific sequence. The AI shows human style from human output, but stores that style as a sequence of numbers, a different pattern for each person. Where did that pattern come from, if not the DNA that made us?
This video is criminally under viewed. I appreciate your input and criticisms. I also appreciate you calling out OPs pretentious bullshit. It’s true that I have a ton of ideas that mean the world too me and not enough time to dedicate to practicing art to make it real. AI has been an incredible asset for me in putting my ideas into reality. Where it goes from here, I don’t know. But I’ll be excited for the future.
Royal, I respect you and I've learned some good stuff from you. I feel like you try to come off as the voice of reason here but this is two sides that are both biased as hell. I think the other guy presents things poorly and sounds like an asshole but has some very good points. You yourself recognize the value of time and effort going into art and creations. A lot of your videos descriptions mention your game and how it took thousands of hours of heart and soul or something. Imagine this future where you instead say "Check out my game it took over 20 minutes of prompt engineering!" Nobody will bother, they will just make their own. As a channel that has primarily thrived on art tutorials, the easier it gets to create, the less demand there will be for your content. I do think there is no going back on AI and whatever he is trying to do to stop it is pointless. But I also really believe it will devalue art exactly as he says. I've generated thousands of images on midjourney etc and it was fun for a little while, then they start to lose any meaning because they didn't actually take any real effort. I'll always cherish my hard -earned blender creations but a generated image anyone could make in 10 seconds just doesn't hold that value. The future of art is looking really bleak in my eyes, everyone trying to be a "prompt engineer" sounds boring and hollow. Personally, I won't give a crap about peoples AI generated art and I wouldn't expect anyone else to care about mine. You think this world where our grand children spin up a michael bay film or whatever is awesome. And I think this world where our grandchildren have no hope of turning their artistic passions into a career is depressing.
Honestly, I feel like most artists, including myself are accepting AI for a Darwinian reason; either adapt to the new tools and new world and thrive in the competitive field of art to earn money or resist and die. Not because we're this excited about it. Personally, I empathize a lot with 2D artists. AI basically stole their talent and rendered them useless seemingly overnight. That is some scary shit. Even writers, voice actors, and composers are now facing that threat. Shit, there are some AI tech now writing essays for students in college. When human effort gets this minimized, it's not a good sign. Who knows what the future holds? Personally, I keep an optimist point of view, but it's very hard to, at least, not be concerned with the threat of AI. Personally, I find the idea of just writing what kind of film you want, pressing enter and boom you have it generated is boring, bland and wack as fuck. I can understand its appeal for NSFW content, but for everything else, human intent behind the work always makes it better.
There's no threat. If it's any consolation, people were talking about GitHub Copilot & Co. replacing programmers but that seems a bit laughable after using it for a while now (though it can help your workflow which is nice). Though maybe programmers dealing well with automation isn't the best example.
Yeah. Even with the same tools, the average Joe with no artistic formation cannot replace a trained artist. AI renders the brush strokes and the muscle memory redundant, but little more.
@@SimplyVanis not really unpredictable, we know exactly how this ai will affect the future of art desu. It's not fear of the unknown, it's fearing about how much tech fucked up the future of art
You're setting yourself up for failure. There is no reason prompting won't be automated just as well. Artists that base their work on their personality and caer are more likely to succeed in this environment than prompters.
tl;dr: "we're at the sweet spot on a very large wave of excitement that will pass. We shouldn't make any new laws until the new-shiny of AI art has subsided" I'll say what I've said elsewhere (haven't watched more than 2 minutes of the video so far, so I'm hoping you cover this): Right now AI art is in a "viral" phase of popularity. There is a wave of excitement that is going to have to pass before we should make any lasting changes to law. Millions of people suddenly have an outlet that makes them *feel* like the artist they've always wanted to be. It can be like someone just discovering synthesizers, they can make all sorts of new and exciting noises, and they are thrilled with results. They see a future of themselves as a popular musician etc. The problem is that this is still the initial "new shiny" rush of excitement. Often times people will drop money (lots of it) on new shiny, only to have it sitting and collecting dust a year later. Because in order to make the new shiny into a full dream takes a lot of effort. Taking AI at it's *current* state, sit down and try to get an AI system to make *your* vision for the image come true, and it's going to take time to get all the details right. Inpainting here, photoshop adjustment there, etc. Again, once people start to hit that "effort" wall, the wave will pass and *then* we should be taking a solid look at the legal ramifications. People who are calling for massive legislation etc. *right now* are more than likely just opportunists looking for clicks and ad revenue, and it would be colossally stupid (or nefarious) of any government to make snap decisions based on what is happening right now. The viral phase of the excitement and popularity needs to pass before making long term decisions. I'm positive that Adobe and other large companies are already looking into ways to monetize things as it is, and with that will come custom models that are "copyright clean", photographers and artists will be offered contracts which will grant them money in exchange for exclusive rights to using their images in models which, coupled with advancement in AI generation tools etc., will all be sold to users for a likely too-steep monthly fee, while AI detection will be incorporated into methods of stifling models that will be deemed "illegal" due to copyright claims.
Glad I commented before watching. Still have to say, I find it interesting how he constantly uses the word "humanistic". He's literally implying that people who use AI are not, in his eyes, human. He needs a solid psych evaluation.
Given how fair use is supposed to prevent people's work being used to put them out of their jobs, I couldn't disagree more. This was illegal from the start, the most important clause of fair use is how it affects the value and market of the original. The idea that we're just supposed to sit back as they pull the rug from under us doesn't make sense to me. I don't see the progress.
2 ปีที่แล้ว +4
Great video! Thank you for taking the time to do it.
But, that is like you pay some one to keep your self safe from other people, but they guy who protects your stuff lets people use the stuff you have. kinda unfair
Here's the thing that I have come to realize. We can put in new laws in our countries aimed at preventing AI developers from scraping art and similar such things as some sort of copyright protection. However, those laws would only work in the sphere of North America and Europe. There are so many countries that don't even remotely care about Western copyright laws, and a cease and desist from us isn't worth the paper that it's printed on. Do they think that a Chinese company developing AI is going to care about American and European copyright laws? How about one from Russia? Or India? Or several countries in Africa and Southeast Asia? Pandora's Box has already been opened, and at this point, would only kneecap our ability to compete in technology with several countries, some of which are our geopolitical adversaries
The most hilarious thing that about that point Steven makes on art in an AI-dominated world not mattering is that one of his other most highly-viewed videos is LITERALLY titled: “You Should Not Care What Other People Think About You Art.” Lmao. His whole video seems like it was made disingenuously and with the sole intent of farming rage-clicks from artists who feel threatened at the moment by AI.
Some folks use art as a recreational activity-they go to the lake, beach, whatever and sketch/paint etc. just to pass time or relax. That’ll definitely still be a thing.
The argument in the end is really toxic, honestly. It's totally fine that you personally agree with companies using your published artwork to improve their algorithms. But you can't expect everyone to think that way. Also I really don't understand why it should be okay for those companies to use your artwork for training, only because you use a cloud service to host them online which has nothing to do with any of that. It's one thing to agree to the terms of service with Google for hosting your files and a completely different thing to agree on third-party companies querying your data to train neural networks replicating your experience and knowledge to sell. If they ask you for permission and you agree to that, everything's fine. But that is not the case. Only because you think that people loose all of their copyright or ownership the moment they make a Google account, doesn't mean that's legally or ethically the case. It also doesn't matter for the whole discussion whether it's technically legal at the moment or not because it's not like there has been a final decision in legislative about copyright violations in content of training for AI generated content creation. So hell no, artist should not be thankful that stable diffusion is providing an opt-out functionality. Because it should be a requirement to have opt-in here. That's the argument. That's the reason he points towards music generation where lawyers tend to be much more involved. The reason why artists are thrown over here and musicians are not is the difference in amount of lawsuits you get by doing this whole training process. It's not about the resolution of the content.
@@asshat8892 What do you mean by my source? I state mostly negatives here like "they haven't asked for permission", "there's no final decision" and others. Have you ever tried to proof a negative? - I thought so. The question is why should training a neural network be treated differently than a copyright violation in case you haven't asked for permission. In the video he brings up that he personally is fine to use services as Google drive to store his personal work. Then he continues that this implies it is valid to use/modify/copy existing artwork from others published via other services on the internet while completely different legal terms or licensing might apply. It's not even difficult to find neural networks being trained on stock images which are typically sold online. So in other words you usually have to pay for usage. If you then train an algorithm to generate pretty much comparable images without watermarks while avoiding its license, how is that legal? It certainly does not fit needs of research. It's just done to bypass copyright. Pretty obvious. Personally I don't see an issue with neural networks being out there to be trained on images with permission by the creators of its source material. However that's not what we see in practice. A little bit of research on TH-cam and you can find multiple videos using copyrighted images, photos, drawings as source or target materials as well. The majority using this tech gives a crap about artists. But then more people might benefit from it overall, right? You could bring the same argument for movie, game or music piracy. As long as there isn't a systematic financial guard for artists, I don't think AI should be allowed to circumvent copyright. Because otherwise who will make the material to train next gen AI? If AI generation is legal, we would actually need to rethink copyright completely. Why would I not be allowed to make a 99% copy of X if my computer is allowed to do it faster? Anyway my main argument was originally that he made a statement in which he implied his own point of view would apply to everyone else... and I think that is quite toxic however you believe AI should be treated. My personal recommendation is downloading an open-source implementation, train it with your own artwork and generate stuff for your own use. In that case nobody should have any issue however they stand on this topic.
I already watched Steven's video, and objected to certain parts, but watching this now brings back the parts which really annoy me. In particular, 44:01. He seems to say that your ideas won't get attention, and therefore won't matter. As though likes on twitter is the only form of validation worth pursuing. His idea of the 'mega-feed' is in some sense likely, since all that's required is the sheer capacity to make the thing and someone willing to do so, but it won't be so all-consuming as he suggests. 'Not even your mother will be able to find it' unless I, you know, show her. Or send it to her. If all of four people in the group chat are capable of producing all the entertainment any of them ever need, and sharing that experience with the other three, I view this as a pretty good outcome. It just means that we must narrow our scope with greater capacity, rather than expand it. The presence of the 'mega-feed' is not sufficient to force you to pay attention to it over a few individuals you actually care about; the only way that could be done is by wresting control of your devices and channels of communication out of your hands entirely. The creation of AI does not make this any more possible than it is currently. That is to say, almost possible in closed ecosystems like Apple's, if not for alternatives like FOSS. There already exist decentralized ecosystems capable of keeping this power in your hands.
I think most of the flaws in his argument stem from his apparent malignant narcissism. Love the piece he's working on... Obviously I have a very limited exposure to him and my assessment is not a complete one, but his language reminds me of every narcissist I've ever known to be a narcissist. Your counters are very well thought out and articulated respectfully, despite your disagreements on the subject matter. Great video!
Artists can be a proud bunch. On artist TH-cam channels that talk about AI, I see a lot of comments that suggest the poster is specifically upset about people using AI image generators potentially getting undeservedly recognized as "artists", while they perceive their own artistic achievements are devalued.
Steven just sounds insufferably pretentious, arrogant, and emotionally manipulative. I can't listen to arguments made by people who are clearly projecting their fears (worse presenting them as valid arguments) and who seem to want more to be an author than a visual artist. As a self-taught life-long artist who is messing with AI, I can say I haven't been this excited about creating art since I was a kid. It is massively inspiring to me. I also have no idea how someone is going to get really unique generations without artistic knowledge. It isn't as simple to generate an image you have in your head as you might think. It's better to go into it without a vision and adapting what it gives you.
Adam duff actually made a healthy argument in a video about where the line must be drawn when it comes to AI, and i think he makes such a great and valid points, i recommend you check it out.
@@done.6191 You know that how? I've heard of many artists that said they aren't getting comissions at nearly the level they used to, and the tech is "in its babysteps"
@@Real-HumanBeing I should be more precise I suppose. I'm admittedly talking about a specific subset of artists, that being artists working to fulfill specific briefs from companies. I'm NOT talking about people drawing furry porn for people or the same Anime Waifu a hundred times at $10 a pop. There is nothing at ALL wrong with doing that, but It was never a long-term job position for more than a very few people. Source: professional artist of 25 years, making money from companies that pay me to do it. None of them have even whispered about AI art being used in any way other than as a tool, which is how it has ALREADY been used for the past 3 years+.
You’re too kind to them. The amount of condescension and entitlement out of Zapata is at critical levels. We’ve gone through this song and dance so many times with evolution in technology, even in the arts. I’ve fallen for it before myself. I used the “canned music” excuse when garageband became popular, and I was wrong. Almost none of these arguments against it are new. As someone who turned to 3D because my hands are too shaky and I’m too busy being a wage slave for corporate to really draw or paint anymore. Pulling AI into my workflow has been so invigorating.
If this really does take off I predict it will either oversaturate the market with CG images to the point where no artist can make money from selling the art and thus balancing the system as the hype dies out and the softwares devolve into artist tools
Interesting statement. I'm a little older, but I've seen the traditional art market become less and less relevant. I think that AI art may quickly make traditional art become relevant again. Things like Gouache, watercolor, acrylics, and oils might be skills that artists want to study.
The argument can be compared to a hiker and a passenger. The hiker hikes for the journey, not the destination. The passenger is only in it for the destination. To the hiker, the destination is the final reward. The hike is a story unique to them. To the passenger, the destination is not a reward. Just something to cross off the list. The hiker will experience the cold winds, blistering heat, and so much, but will be one with the world. The passenger is protected by the carapace of the vehicle. The hiker can arrive at 10 destinations and will feel fulfilled. The passenger can arrive at a hundred destinations and will never feel as fulfilled. The journey has a destination, but is it really worth it to cut the journey out to instantly get gratification that might've been more fulfilling if you hiked the road yourself? You are a passenger, you are in it for the destination. And the vehicle is too fast to fully enjoy the vistas. And if AI is to be a vehicle, it is the fastest vehicle. So fast, that the only window in the front shows the destination. The journey is negligible, the vehicle is too fast for the journey to even matter. I love your 3d tutorials for their destination oriented methods. But if it is about art, it's better left with the journey so that the destination is fulfilling. Calling the journey a waste of time, is a waste in of itself. Your philosophy on getting to the destination fast feels like it only applies to coding and technical stuff,. Like they are actually guides on quickly learning how to "walk" in order to go on a hike. But the direction you are going... Nope, I will not follow you to your super fast vehicle because I want to enjoy walking and you have way too much strawmen. The draftsman show recently started again with this topic actually getting professional voices on it. But say what you want to say, I think AI collages makes human art more important than ever. Like, "huh, this is made by AI?" and "wow, you didn't use AI on this? Amazing!" Edit: I still love your 3d and coding videos. I think you should just stick with them.
Yeah some of his arguments just infuriate me. I am not an artist and a philosopher, but I am an artist and a philosopher. I am not the former because those are not my professions and I don't hold degrees in those fields and I don't make a living off of them, but I am the latter because I have created new work from my imagination in various ways and have pondered and discussed the mysteries of life and existence. Just because I am not publicly recognized and money is not exchanged doesn't mean that my work and thoughts can't have value, to me, my friends or some random people on the internet. I know some people that feel they can't write the stories they want to write not because they think nobody will read them and won't make money, but because they can't live up to the standards of some random people on the internet and because they fear those who complain about everything and generally as groups are vocal and have opposing views on various subjects, so they are stuck because they care too much. That frustrates me a lot. Why should we care about what others think or want in order to create what we want to create or how we create it. The elitists can just shove it. Why should they belong to some privileged class that can't be disrupted by new technologies and other events, everyone else is exposed to this risk.
There's one thing I think you may have misunderstood in the section at 45:00, your biggest disagreement. Steven said not that the dream anime is worthless, but that in a scenario where art is effortless, and therefore the supply is limitless, that no one's art will be able to get a foothold/a strong share of attention.
You won't believe it but even within human only competition most of worthy art doesn't get attention it deserves, and in cases where it does get attention it's mostly because artist already has established their brand over the years. Also limitless supply of art is really a pro and not a con of this situation, unless you're 1 percent of population that required that to not be a case to extort people of their money.
@@xn4pl I'm not sure doing a job to try to feed yourself counts as extortion. People's jobs are being upended, but their job isn't an extortion racket, anymore than a plumber fixing the pipes in your house and wanting to be paid a living wage for their effort is extortion.
Watched yours, Jazza's and Shad's videos regarding this topic. My concern with AI art has never necessarily been that it "steals" art from artist. But rather two different things. To keep it brief; rather that as this technology improves, and its used by major companies or future "artists" the unique aspects of artistic mediums will be boiled down to an algorithm. Things will become "samey" because that's what people like as a whole. As someone whose hobbies tend to be artistically driven (yes much of it using computer assisted tools, yet still deliberately chosen by the creator) I don't like the idea of future media being driven by a formula and the potential homogenization that will bring with it. Not only that, but its hard for me to see AI art as something that was "Created by an artist" in the same way. It would be like me paying an artist for a commission, asking for different tweaks and changes as the piece progresses, and then claiming the final product as "my art". Yes I may have the ability to perfectly envision what I'd like the image to look like in my mind, but not the skills to manifest it. Telling someone who does have the skills to create it all the minor changes I want them to make until its made manifest does not make me an artist, but more akin to the "idea guy." The artist made the work, I told them what to make. AI created art, in my opinion, follows a similar path. I'm sure theres small uses of AI to improve ones art that I'm okay with, such as maybe tweaking a pose, or coloring here and there, but there also comes a point where I don't really see it as "you" doing the art anymore. I'm not sure exactly where that "line" is just yet, but working on figuring it out. I would not consider myself an artist however so maybe my opinion is off base, but thats how I see it.
I come from Shads video and i actually agree with you on the latter paragraph. I consider myself a digital illustrator, and im currently studying graphic design. I currently use Photoshop to do my art pieces because of the number of tools it provides: transforming an image, levels of brightness, the crtl+u that opens the Hue/saturation/brightness tool and allows me to twerk stuff quickly... Its amazing. And i do see AI´s enhancing my work. EG. I had issues this year because a stock picture that i wanted to use for a work of mine wasn't long enough, and i needed to "invent" the rest of the image by using tons of collage, tampon and my graphic tablet to fill and make the image seem more coherent. If an ai can do that tedious job for me, im all in. But when it comes to art, i still feel Ai´s although legally allowed (because we all had to suck the terms and services of any social media) aren't morally correct yet. I´ve seen Shads process when he did his wife, and the Armour design isn't his making, and if it were me i would not claim i fully did that piece of visuals. It feels like a cheating way bigger than the ones i could do on photoshop. I wanna embrace ai, but i fell so nasty for using a bot that serves me everything done
As a person who suffers from aphantasia, I have always struggled with lighting in painting among other things. I never post an AI generated art piece on my main page and call it my own, I always work on it quite a bit. Once I have the lighting figured out, I usually draw over it, making it my own. I am not against AI but it does need some safety protocols like maybe you should not be able to generate realistic images of regular people and put them in a crime scene for example, this scares me. Using it propaganda pieces, I dunno... not everyone out there has the best ideas in mind. Look what happened to Facebook and Twitter, people now use it only to spread hate and propaganda. I fear AI is capable of making things worse. I do enjoy the process of ideation with AI, its given me ideas that I would have NEVER thought of, otherwise.
With a condition such as yours, I totally empathize with the usage of such AI generations. In fact, I think it's so important to have these conversations about artist's rights around AI specifically because of people in your position. The current AI models have a large number of unanswered questions about legality, ownership, and what level of input by a human could place it under the "Fair Use Under Modification" clause. For me, that's one of the main reasons I'm so against this narrative of AI Vs. Artists that is being perpetuated by both sides, despite both groups claiming that "WE'RE THE REASONABLE ONES, IT'S THE OTHERS THAT ARE MAKING THIS AN ISSUE". With licenses such as CC0, CC 3.0, and of course the plentiful number of items in the public domain, there's a clear-cut way to make an algorithm for people in your exact situation without having to even ever touch a website like ArtStation. Typing in "Mario Kart by Simon Stalenhag" or "Seinfeld in the style of Banksy" is fun, but when you start commercially selling those products with the intention of mimicking that artist, then suddenly there's a big free-for-all at the fourth clause of Fair Use, and everyone suffers for it. It's not about removing the AI, it's about making sure the AI protects both the artist whose content it's trained on and the artist who might use the algorithm for inspiration.
And that's how I think AI generated art should be used. Sadly many people don't understand that or utterly missrepresent/missread copyright laws to not think about how AI assemble art
@@masterzoroark6664 It's not about how it assembles the art, it's about the fact that a machine (one that's going to do a way better job of learning patterns and reproducing things because that's all it knows how to do) is being trained on people's images to learn. It still has to use the image at some point in the process to learn, therefore at some point it becomes about unauthorized usage of images for commercial purposes, thus the argument still needs to be had.
@@mbishop3d832 yes, I know and stated that, sorry for missunderstanding. We just probably need to wait till one of these generators spits out pracitcally Disney or Games Workshop merch for people to really notice they can't hide behing "Machine did it". Tho on the artstyle point- so far AI showcased it's unable to make a style of their own. If you ever drew or saw someone drew you know that with each attempt to draw like your idol you still will draw differently. That's how artstyle is formed, and only disingeniuous forgers don't ebrace this fact. I am of belief that once AI makes artstyle of it's own and artwork from it's own creativity then the artwork is it's, not the person's who overseen it create.
As someone who regularly needs art commissioned for my musical projects, I've hired through places like Fiverr and artists I've known personally. With the exception of one artist who I could literally walk down the hallway and look over the shoulder of, almost all of the art I've ever commissioned was discovered to be photobashed or outright stolen, despite paying hundreds for it. I was able to backtrack through and learn that EVERYTHING was always photobashed or straight ripped from other copyrighted sources, even though I had paid for commercial usage rights and spoken directly with the artist. In my attempts to support the independent artist communities and paying multiple hundreds of dollars for individual pieces, it's only through AI art do I actually feel SAFER from copyright infringement and like I'm actually realizing my vision. There's also an argument to be made that for musicians like myself who operate on very rapid release schedules, being able to sit down in a single day and get the art I want and pair it with my release is a tremendous value for audience, as I can deliver faster and with a more compelling image to accompany my music. I can say with confidence that AI art has never replaced a commission for me though. I would have made the art myself (like I had for years), but now the quality level of what I can do is far and away superior.
You make a good argument for AI art. I'm sad but not surprised to hear about the rampant photobashing and stolen content artists are selling. Fiverr may be part of the problem. Artstation seems like a place for more reputable artists but wouldn't work for inexpensive, fast release art. Frustrating but a good case to learn AI for art.
@@The3Dsmash Fiverr definitely contributes to the problem, since it rewards fast, cheap, and easily produced art. It does feel like it's a "pick two: fast, cheap, or quality" type of deal. But on the other hand, I've hired artists for illustration of comic books and had a far more pleasant experience, so it seems to be very much a crapshoot, haha.
To be honest what he finds as a horror at around 49:00 I find as a dream. Where it finishes my lost or trailed thoughts. Is a helping hand getting past my creative depressions. Ends up being a tool to help change my perspective on life while still being tailored to me. It would be so cool to have ai movies generated just for me to consume so I don't have to wait months for the next A24 movie to drop which already seem like movies tailored for me. I think it is crazy to believe we wouldn't want to see a world that revolves around ourselves as most of us already seem to do. Human interaction should be a choice and not mandatory. Replace workers with robots and machines. Art with generated components be it filtered through meat or metal. Creating the next step of humanity and making something more than human is what is (in my opinion) the natural progression of the human condition. @Royal Skies. I love your mentality and that you (like me) believe we should make things easier for others. That just because you spent so much time on something doesn't mean you shouldn't give someone else a short cut. Plant your trees even though you might not ever live to see their shade.
@@TheCaptain610 first of all I run off the theory (which I have yet to find anyone else talk and is something I'm building) what if God built us to be better than him obsoleting himself. Granted being an atheist I counter my own theory but it is still in the works. Though nobody would be billionaires because like in our world no money is completely baseless though would actually have zero need. People who aren't in that upper class range if money does still play a roll in the economy would do whatever. Yes robots do everything but it doesn't keep people from doing anything. People can still pursue arts, maths, sciences, or stem jobs like they do now. Just there isn't the need. It is a want. That's the biggest difference. And something I think literally everyone overlooks in an automated world.
@@TheCaptain610 Whatever a person wants to set themselves as. If UBI becomes a thing, I see it simply as the ability for people to find what they truly want. This can either be someone whose nothing more than a consoomer, or someone who decides to explore and become knowledgable in very extreme niches. The ability of people to be able to live for their own desires without a job taking up the vast majority of their lives just to put food on the table is a net benefit for humanity.
@@TheCaptain610 I didn't really elaborate on what my ideas for how to make UBI would work, which I apologize for that. My simple idea is that the gov would have nationalized space/moon mining that they would use the resources from to fund the UBI. I do not trust the CEOs at all to do it. Of course, this would require the government to have some actual changes to address the corruption currently in place to actually facilitate this (boomers are still in power after all), but I think this is the most realistic way to achieve UBI, and I don't think the government will just get away with trying to let mass amounts of people starve as automation becomes more widespread. It's one thing to have the majority of the country living paycheck to paycheck, another for them to be unable to eat.
Man, I follow you for about 1.5 years probably, started from interest to BNPR-render-blender stuff, and really like your attitude and style and determination in previous videos. But this one come out too protective, which means yourself was kinda emotional. There were some slightly better arguments towards the end of the video, but I'll say about appealing to Google and other's big tech Terms of service is transcendental level bullshit. It is kind of way of saying: "Well you uploaded your pictures to internet and taken photos with your phone, they'll be used and abused anyway, feck you bro." This is bringing up immoral legal loophole, which should not exist by common sense, but is present to serve corporations, because they are and always will control legal field to their favor. Also I am a professional 2d artist and at recent times used Stable diffusion super heavily, and got awesome and diverse results from it. And this did not made me happy at all, for variety of reasons. (allright, nobody cares about my tiny puny artists feels, no problem with that) But we as humans are loosing focus from real quality, struggle for personal growth and improvement, to quantity and superficial approach. We get more and more possibilities to do stuff, but games and movies for weird reason do not become better, quality deteriorates. probably around 80-90% of top notch high quality art from artstation, for example, is already looking the same and boring. Everyone is repeating everyone else, and now AI comes to play, and for now it is awesome new toy, yeah, I enjoyed playing with it too. But over time people WILL lose their personal voice, and only one who will benefit is provider of AI. Also call me the tinfoil hat guy, but I do think that going opensource was a smart-ass marketing move, no more and no less. And also you know what, I would love to be wrong on this, and I would love that Steven would be wrong on this, and everything will just become better with it. But knowing life and dem peoples, most probably it will not. We'll see.
I think that AI will replace artists, because why should you pay an artist if you can get the desired picture for free. 2D artists are already almost completely useless. (only animators are still needed, but not for long). And the bullshit that these AIs will serve as tools for the artists, and that the artists will be needed to work properly with the AI, is nonsense. Artists simply won't be needed, because it's incredibly easy to learn how to work with AI, and the scenario in which a person will be paid to enter a description of the picture and click the "generate" button is just as stupid as possible. The truth is, that artists will very soon lose their jobs and that's it. No alternative jobs related to working with AI. No happy ending. There is no longer any point in artists job. Everything is simple.
AI cannot replace artists, if artists become extinct in one day, then AI too. Because AI will not be able to train without artist work on the internet. Do you know Vocaloid? The company that created Vocaloid hires voice actors to and stored their voice in a voicebank and then makes a synthesizer that can sing and make music based on the voicebank. Did the singers lose their jobs? No, they didn't. In fact, they got more work because of the trend of covering Vocaloid songs, which is intended for people who like vocaloid songs but don't like the mechanical robot voice. And now many music synthesizers use AI Voicebank, in order to get realistic tune like humans. Did many people lose their jobs? No, they won't. Instead, many people who understand music but can't sing will become music producers using the synthesizer. And AI is a tool for artists to improve, especially since AI images are copyright-free. Artists can improve faster with the method of training by copying without having to fear being sued and made into drama on twitter
I don't think it matters whether current AI training sets contain little to none drawn images or whether it is mostly based on photographs. It's still violating copyright as far as I can tell and the set of weights for the neural network couldn't be calculated without the data.
I fundamentally disagree with your vision for the future described at 6:20. “Harry Potter lightsaber red ranger adventure” or whatever already exists, it’s called a Content Farm. “Among Us vs FNAF vs Friday night funkin at Squid Games Spider-Man Elsa etc etc.” and plenty of iPad babies are already frying their brains on it. You call it awesome, but I don’t think the whole of society’s entertainment should be the cinematic equivalent of jingling colorful plastic keys. Even if AI got better than that and was able to make content that seemed “meaningful” on the surface, that is still largely by accident. I would rather experience art that has meaning on purpose. Your bullet hole analogy ironically shoots your argument in the foot in this regard. Then there is the ethical dilemma of how many people would be put out of a job in Hollywood if studios starting implementing these technologies and replacing employees. Suddenly thousands upon thousands of people are out of a job in an industry killed by AI so they can’t apply their skills for gainful employment anymore. In the ideal future presented by this video, a slew of struggling artists and unemployed writers and animators sit in their decrepit apartments as society looks on at Content Farm content presented as mainstream, devoid of the humanity, soul, talent, and hard work that makes art worthwhile at all. That is Hell.
You should watch the movie "Tim's Vermeer", A technician analyzes Vermeer's work, and figures out how to draw a Vermeer painting without any painting skills. It shows that even old masters wouldn't hesitate to use a new technology to create their art.
The only argument you put forward that I really have a strong disagreement with is the concept of consent. Where terms and conditions that are written full of legal jargon that would take even the best lawyers time and focus to pick apart, the fact that most people hit 'I agree' does not, and should not, automatically mean they consent. A company cannot for example just insert a clause stating that by using this software you agree that the company can at any point in the future demand 100% of all the revenue you made from using this software and expect it to be upheld by the law.
I use AI for writing because I have limited use of my hands. They are my ideas and I am directing the flow of everything. I still do a ton of editing, organizing, fixing and rewriting though. It's still a ton of work. When I had use my hands I would have never used AI and been on the side of the artists going against AI. However, after I lost use of my hands I thought my life as a author was over, it took too long to write anything. AI has helped me get back into the saddle and my life has meaning again.
It’s important to note that even though the current data collection practices are technically legal, copyright for AI generated art isn’t. There is already precedent for this, so as long as the law don’t change, artists like myself just cannot use it commercially and professionally at all, all you can really do right now is generate and share on social media, which will make your art and it’s idea open to plagiarism without much legal action available to take. Not to mention stable diffusion right now still isn’t robust enough to generate consistent images, it really struggles with recreating the same designs at multiple different angles, without changing the fundamental design all together, and it also doesn’t allow you to generate images in layers yet, so you also can’t really change the lighting or local color without again changing the whole design all together. We need better AI design that give you more control than what we have right now, that’s the only way for it to truly become a tool for artists instead of a cheap way for normal people to generate random images, because artists don’t care how good looking the final result is, they only care if it can be easily controlled by them and create the way they intended, instead of an accident, and robust photoshop layers, clipping masks and blend modes with strict mathematics behind them will always be more robust than neural networks based on probabilities generating images on a single layer.
One of the best videos of the year, SuperEyepathWolf's "Influencer Courses Are Garbage", he talks about the joy of creating. About expression for it's own sake, and delivers the powerful message: "There is nothing wrong with you if other people don't pay attention to your art. That art is for you. That is enough. I promise." According to Steven though, your art is worthless unless people notice it. I am not going to attack him for his views, but that, his elitist, gatekeeping attitude, and his constant appeal to emotion are REALLY rubbing me the wrong way.
Despite our many disagreements, I wanna go ahead and say that I believe Steven did a great job presenting his points, and that I wish him all the best with his endeavors for applying pressures where needed - Please be kind and respectful to him and his community when you visit his channel and watch his original video here:
th-cam.com/video/tjSxFAGP9Ss/w-d-xo.html&ab_channel=StevenZapataArt
Also, shout-outs to Jazza for the hilarious Fiver video on his channel which can be found here:
th-cam.com/video/5GO2xKmZsVo/w-d-xo.html&ab_channel=Jazza
If you're interested in creating fun art and installing Stable Diffusion locally on your computer, you can checkout my video playlist explaining how to get started here:
th-cam.com/video/6MeJKnbv1ts/w-d-xo.html&ab_channel=RoyalSkies
As always, hope you have a Fantastic day, and I'll see ya around -
@@ScottSpadea got flashback Ethan moment from Cod infinite lol.
@@Christian___ Steven ain't arguing, he's gaslight the people who don't agree with him.
I think you should have an actual debate with this person instead of making a "reaction video" (I'm not being sarcastic or condescending here I just don't know what to call it), as this method gives you the last word on pretty much every point making this all sound one sided. While I agree with a lot of what you said and you are still missing and/or didn't address some really important parts. Many of which come from your own perspective as a successful artist.
@@ScottSpadea
You made several interesting points, but I'll just raise one: The AI is not intelligent, it is a flawed program written by humans who admit they don't understand how the program produces its results. No program ever written is perfect. No computer part is made without a mistake. Add these up, and throw in a data-set of information from people who hate, despise, refute, ridicule, belittle, and reject real art for their own personal pleasure, and the AI is going to spit that out, more or less, in every image. The problem is the human condition, not the technology. The motto of the computer scientist is this: "Garbage in, garbage out". Let that sink in before you decide to take a trip in an AI vehicle. Love the technology, be cautious of the people who produce it.
Wow, I don't know where you got the link from, but I did send it to you, glad and surprised you could take it that seriously - from 90 seconds to 90 minutes! Is that a record? I actually don't think you addressed the points raised, although you went through them, so I am still not sure what to make of his arguments, but at least you've clarified your position. I'm happy to go through all the points, but not sure what format to use.
I believe you both have more in common than you think, but it's like describing different parts of the same elephant. You both want the elephant just don't agree with the part the other person is holding on to. Personally, I'm not into AI nor Art, but what I understand about both is that the programs are blurring the lines between copyright law, intellectual property, and ownership of a work. It also makes your definition of art - which I accept - difficult to apply to a program, which is what AI is.
So I don't have a problem with the position you hold, and it is probably a positive result for the 20-30 years AI researchers received mockery and ridicule for getting no-where. Whose laughing now? (ssshhh.. I still don't think it is intelligent...😉)
That said, some of the responses didn't seem to touch on the thrust of his point and I'm not sure if I missed the actual points because I'm not an artist, or that his argument wasn't addressing the actual issues. And that's what I'd like to resolve, if possible.
I swear, more half of Steven's video is spinning up an "Us vs Them" scenario using the most emotionally charged terms and constant assumptions of malice. Its so frustrating trying to see his point of view when the entire ramp up to his point is inflammatory fluff, the crux is ignored for a single aspect, and the payoff is condescension and calls to action. Distilling some of these talking points could be summarized as "you want X? you don't want X. You're bad for wanting X" with only hyperbolic rabbit-chasing to back it up. Royal, you did a fantastic job pulling the most complete interpretation of his points out and addressing them fairly.
And for all those years you spent learning 3d modeling, the grind of working out code problems, and dealing with arrogant cockends who seem to get all their sense of self-worth from internet forums, I can't tell you how much time your guides have saved me. Whatever your process is for trimming out unnecessary information and compacting your presentation is, it's absolutely worth it from a 3D hobbyist's perspective.
Shad was right this video is criminally under viewed.
Nope, the video from Solar Sands is under viewed, everyone should watch that before making a final conclusion on it, it is not just simply about new technology and full stop, it is very complicated
@@jaxkk1119 that was a good video thanks for the recommendation
Watched the whole thing, and I've been thinking about it. I think i decided that no matter what my opinion is, I respect you greatly for doing research and keeping a level mind, and generally not being a jerk either for or against AI art. I'm not sure yet if i agree with you on everything, but I definitely respect you. Thx for putting in the time and effort
Hey man, thanks. I'm the guy who sent this video to you on twitter. I appreciate your take on the issue.
I remember watching part of this video some months ago. I agreed with some of his points, but i didnt finish it because the narrative is clearly and sharply based in fear of the unknown.
I sure hope ai isn't abused by or for corporate profit margins, but i definitely think everybody artist or not would enjoy or benefit greatly from these types of advancements.
I'm not an advanced artist, and i havent spent much time or have time currently to learn more on blender, but i can see indie game devs getting a massive productivity boost from ai advancements. Just like how 3d modeling software advanced in the last 40 years.
I'm glad Shadiversity recommended me to head over here. Suggested a friend to watch your playlist on Stablediffusion and appreciated it. I'll give a look at that playlist as well since I think it can probably help me visualize my ideas from descriptions of my characters and create something cooler than what my words describe. Thank you for making this response video.
Technological progress will never be stopped. We know this simple fact that has defined humanity for years. The printing press made it far easier to mass distribute writing, the typewriter revolutionized writing large amounts of words like novels, the automobile revolutionized how we get around, and the personal computer changed how we live our lives every day. And now we have AI which is poised to revolutionize how we can make art
As a computer engineer that has also worked on AI, thank you for clearing up misinformation. I dislike elitist people like that dude, half of his points were "I'm a real artist" basically, not to mention that people like him forget that AI has been available for years already and in far greater and more useful fields than drawings, like for example in medicine, but you don't see doctors protesting, they actually love its help.
a calm and direct response towards what I can only describe as pretentious, presumptuous and sophistic rhetoric which sadly a lot of artist are very easy to fall prey to
This is sorta out of context, but 35:39 this story really hits me.
I've had the same experience reading trough programming fourms and people just refusing to answer (most recently with linux).
When I was starting out in blender and I had the dumbest of issues, your channel was so informative and so straight to the point that not only did I fix my problems, but I binge watched the rest of your videos and learned so much from it.
Things that I would have waisted hours of my life because I forgot to check a box were now something I knew how to fix and not only that, if I forgot it or if it had some small differnces, I always had the option to check your channel and watch the super quick video again...
IDK if you get this often, but thanks for steping out and being the opposite of the fourm guy in the 35:39 story.
I've been coding for years and usually stay away from Stack Overflow. The only overflow in that forum is the amount of c**ts.
I remember as a kid I tried to make Minecraft mods and people would just say "Learn Java" instead of helping with the issue. When I got older I relized they are just assholes who know enough programming to use it but not to teach it. I'm pretty good at programming now but honestly you could learn a decade of my experience into one year nowadays and I'm happy about that.
i wonder how that guy feels about schools; i bed he'd make a greeeeaaaat teacher.
student: Mr.shitbag, how do i do #5?
Mr.Shitbag: that's Professor shitbag to you, and if you wish to know how to do it, you will have to figure it out yourself, or go to a 4 year college like i did, where a professor taught me everything i know.
The story at the timestamp you mentioned. It hits me too.
so we're in agreement on that point.
But when we're talking about AI-ART, most artists don't argue "it took me years to achieve this, its not fair you can do it in a click of a button" They dont, not in terms of Ai-art. Because "at the touch of the button" has been there for years, in things like 3D and lighting. You don't need to understand lighting, you just need to place it etc.
When we're talking about Ai-Art, the argument is "it took me years to achieve this, and its not fair that you stole this work from me to / used my work to train your Ai and call it yours"
Thats why most artists are ok with Porgrams like Adobe's Firefly, that uses only paid-for/licensed images to train their Ai.
I am having almost the same issue on my project. I actually need his lua script on how he made the enemy AI work (making the enemy, smartly, jump out of the water towards the land and hunt you down especially if you're not on the water. They have no issues jumping down the water though.), but to no avail. I even warned that there will be jank on my enemy since I couldn't figure out how to make that kind of script nor get to know the secrets on how he made it. I keep asking this for like 10 times already only to ending up with a straight "no" to the point of me going to the extreme that I am already harassing him that I really have to stop nagging about it because I am getting too far on my methods.
The TL;DR is. He's not willing to give the script nor tell me the secrets, not even a single tip or, hint on how he did it to the point of me getting a bit pissed off and going a little bit too far.
Love hearing your thoughts about art, you are obliviously thinking very deeply about this stuff, its a pleasure to listen to!
The part about the imaginary anime made me hate him. That was a disgustingly arrogant sentiment.
It made me actively hope Artitron 2000 takes his job
I agree with you that the narrative AI vs. artists doesn't help at all. But I don't think that's the point. The narrative he seems to make in the video is artists vs. the companies selling AI and I think that has a point. Because Google won't come to you, asking for your images to automate the process for you. They will either charge you or automate it for themselves. I know this sounds stupid because especially Google is mostly a service provider currently. But the difference is that while they only provide service now, they can provide the product (art) themselves later because of the AI training you agreed on. Because with your art, you are giving them your personal experience for free to feed their algorithms. You experience is the product here and you are giving it away for free or far too less, I'm telling you.
Of course those companies won't start doing that tomorrow but they will get there. Now they are all still improving the algorithms. So it makes more sense to allow others using it as long as they feeding it with data. But as soon as they cut the line, your career as artist is probably over. With more and more digitalizing, I don't think physical art or "human art" will make much money in absolute numbers either. Sure, it will work for a small amount of artists being kind of celebrities because that's how premium branding works. But most people will take AI generated images over manually crafted artworks anytime because the density of content is higher, details will be finer, resolution bigger and cost much lower.
@TobiasFrich "They [Google] will either charge you or automate it for themselves. "
I think you're on to something, but not quite there. The point of Google's ToS is to transfer your copyright to them so Google controls the distribution and sale of your work, not you.
And this means any company who copies your work into an AI dataset is violating Google's copyright because Google doesn't allow it. And this applies to all companies that require you transfer copyright to them. They want control over your material for their profit, not other companies, and certainly not for you.
So whether or not you signed away your copyrights, the work is still under copyright law, and has a legal copyright holder that can determine the conditions for copying 'their work'. They will, of course, freely allow anyone to copy your work, for a price. And that is where the dataset company does the double take - "we're a non-profit, for research entity and therefore not-for-profit and can legal copy your works under fair use, without payment." After which, the dataset is moved to a for profit company, completely legally which they describe as a capped entity (or whatever). So, now Google's attempt to profit from your work has been wiped away by these entities because they can 'sell' your work to other companies at a 'discount' compared to Google and every other company who got your copyright. ooof!
@@petergostelow To elaborate on a key point: They have rights if you upload it to a Google service, not if it's automatically added to their algorithm. That's the key difference. If I make a website and Google takes the images off it, Google doesn't own the rights to that image.
They even say it themselves: "Google does not own the images found via Google Search. The "Usage rights" Search tool is provided to help you find images which may be suitable for your use."
I'd post the link to the article if TH-cam didn't filter my original comment with the link as spam, haha!
I'm sure if you search the quoted text in Google, you'll find the article, though.
Google has absolutely zero control over images in the search results unless you manually upload them to one of their services!
@@mbishop3d832
You're right, although I wasn't referring to the search engine, neither was the video. Somebody always has copyright, even if it is public domain. So anyone who randomly copies an image could well be violating copyright law, whether from the search engine's results, or your website.
@@petergostelow Ah, interesting, could you elaborate what the video was talking about then? I'm a bit confused now!
@@mbishop3d832
Yes, it was with reference to 7:28 'counterpoint to stolen art'. I am just pointing out that all works have a copyright attached to them, regardless of the Terms of Service because the ToS does not remove the copyright, it merely transfers the rights. Thus, the ToS is not a counter the theft claim. It's true they're not violating your copyright because you no longer have one, but somebody does, and that's why companies have created the work-a-round capped entity. The 2nd fact is, that the companies have spent large sums to create a specific economic entity to download, should alert you to the probability their lawyers know something we don't. So the ToS is irrelevant to the theft claim, imho. Did I miss something?
I want to point out that - "Web scraping is completely legal if you scrape data publicly available on the internet." meaning they don't even have to get you through the terms of services of other platforms, and then buy the data second hand, or second rate. They can kick it up, and just get your data directly from the source no middle man required. It's not theft if it's legal. You may not agree with the law. But when you use legal terms such as theft, the law has to be taken into consideration. And considering the laborious processes surrounding AI Art, there's a strong case it falls under fair use. Which would grant a level of legal immunity, if granted by the courts.
If your art is private, it's not protected from viewing. AI does not cut up and rearange your image, it views said image. But it doesn't use the image itself. It use the data it receives from said image. And creates based on its programing it's own image. The issue lies with the multitude of images. No 1 image makes a single AI image. Millions upon Millions images (data), related to the prompt are used to create a new image. In that sense, it can very easily be argued as transformative. Because 1 image does not lead to another, it's entire dataset does. Meaning it's essentially drawing "inspiration" from all corners of it's databanks. Thus making it impossible to draw a directly line from 1 image to another. Unless a separate method of AI Image Creaton is used.
12:07 yeah this is a pretty good summary of 95% of all internet conflict. The "us vs them" mentality is simultaneously very enticing and very destructive, being a carryover from the stone ages and all. Zenata is purposefully getting the audience riled up and angry because anger bypasses rational thinking, and makes it far easier to draw in onlookers that would otherwise be neutral. It's rather insidious
This is not a debate, this is a video response. A debate would be that you and steven explain your ideas in a direct.
I am an artist and have messed with the AI programs, I can say that the machine by itself will not be able to replace an artist unless (ironically) and artist works the machine.
Not to be rude to Mr. Zapata, but though his argument is good, there is too much emotion and fear-mongering covering his points up.
Also, you could change the topic of "AI art" to something like "game engines" and the argument would sound similar. Though I do know bad people are going to abuse the generators, I don't think scaring people, especially artists, from using the programs is going to resolve the problem, it actually might make it worse.
Edit: NVM Zapata completely lost me on the second half when his arguement became an _"arguement"_ ... well, back to drawing and rendering with Stable Diffusion at the same time.
Good Vid Royal Skies, glad Shadivesity directed me to ya :)
It certainly doesn’t promote workable solutions to whatever problems do exist.
About the feeling argument at 1:02:00
There are a lot of people that enjoy growing vegetables in their garden. By the "artist" logic we shouldn't mass produce vegetables not to hurt their feelings
i think his point about your art no longer mattering is more a point about making it even harder for artists and AI generator users to turn their passion into a living, which is a valid point since there'll be a billion more people flooding the market with their generated images
Ehh, as if there aren't a billion starving artists competing against each other already. This just proves even further that without sheer luck, art is not a viable career path.
@@DonVigaDeFierro oh it's definitely not a case of just luck currently, if literally any potato can just type "gimme pretty picture" into a textbox it may well become that way though
@@tieguaili3d298 While "gimme pretty picture" is pretty funny, I'm genuinely tempted to put that into stable diffusion, I'm curious as to what will happen!
@@mbishop3d832 yeah i'm gonna need to try it out as well see what happens, no other guidance to the prompt, just that
ANYONE can draw. Not everyone can draw well enough, or train to get the aesthetic sense that people will PAY for. A billion people making shit are still people making shit. AI is not capable of, and in the near future will not be capable of, producing unaided to a specific brief. Now, if one's entire living is based around making pretties as fine art, well...that might be a tough sell, much as it is a niche market for artisinal glassblowers (which still exist, despite glass manufacturing being automated to a large extent).
they're not automating the boring shit though, if they were they'd be automating retopo, UVs, rigging, technical bullshit that takes forever and no one wants to do themselves cause they're rather be sculpting painting and animating, but instead the AI people are automating the fun creation parts not the boring technical prep parts
THANK YOU! This is what those ai programmers don't really understand. If there was an AI for retopology, that would be amazingly useful. Currently AI programmers are doing the wrong thing, they should really consider making things like you mentioned (AI powered retopo/UV plugins) instead of sapping the fun out of creating stuff.
@@quadrantthree static image generation was easy pickings for programmers. There's already a lot of algorithm based retopo and uv based automation. But it's a lot harder for an AI to understand it for example texel density and specific layouts in uvs are often frankly more subjective to the project or task. Than feeding a bunch of results into a model for replication since ideally all results fed into the AI were generally the best examples of what it should do. When all the information is selected by us then any result given to us by stable diffusion will be good basically. The moment you ask it to do something it hasn't quite seen you'll begin hitting roadblocks. This is why stable diffusion video is trippy currently. It'll take 8 more years to fully integrate AI into 3D and 20 more years to integrate AI into visual effects and engine programming/coding. And that's if everything goes well
You have no clue what is hard for AI to do. What you think AI scientists and engineers should do is irrelevant, because you don't know shit about the field. We're building intelligent system from the ground up. This type of technology, text-to-image models, weren't created with artists in mind - they're all subsets of natural language programming, computer vision, and, with diffusion models, thermodynamics.
The fact artists got hit first is just an unfortunate result of what was discovered first. Artists may have always been the first to be hit, just because the visual components necessary for art is just in line with computer vision and image generation.
Unlucky. But dont tell me and my colleagues "what we should be working on". We've built this field from ground 0. We discover things over time and automation follows from our findings.
@@leamJG how bad does your butthurt have to be to go off on a rant like that because someone pointed out that your discoveries are of no use to anyone and you'd have less people trying to get your shit shut down if you looked outside your little bubble?
@@tieguaili3d298 discoveries are of no use to anyone?
You know AI is used in cancer research and in the health field, right? We can identify cancerous tumors better than doctors can now. We can accurately predict and simulate epidemics. Our work on neural networks have helped neuroscientists learn more about the human brain.
Get your head out of your ass.
“You want it to matter you told your story.”
I mean, what? I don’t sell any AI art, I don’t post anything, I just show stuff to friends and family who are interested.
The fantasy warrior with armor that looks like a jester with flaming swords matters to me because that image was in my head for years. I could care less whether other people validate me, or if they think it’s stupid, it matters to me that I can finally see it.
Those who think that humanity now has that ability have their priorities completely out of whack, they care more about praise and validation than actual imagination.
(Edited because I duplicated a words)
I think what ticked artists off the most was the creator of Stable Diffusions comments on Artists "My message to them would be, 'illustration design jobs are very tedious'. It's not about being artistic, you are a tool". That rubbed everyone the long way.
They make it seem like it's tedious when every brush stroke is a blast and instrumental to an artist's well being. It's not about the final picture, it's the journey there. The most popular art in history is unfinished.
This is going to replace FIVR artists if anything first.
I actually fall into Royal Skies' category of artist... I enjoy the end product, but find the process very frustrating and tedious. Different artists find different aspects of art fulfilling. Mine is in getting ideas out of my head, and onto a computer screen (or onto paper, as I like to sketch ideas quickly sometimes.)
I'm all for AI art generation tools. I specifically like the image2image function that can take my original concept art, and then iterate on it, adding details and such that might inspire me to add other things and change it around. I also like the functions of inpainting and outpainting to fine-tune certain parts of the image, allowing far more control of the results.
These tools have re-invigorated my desire to work on concept art, whereas before, due to the tedium and frustration, I had been suffice to just imagine them for quite some time now. (as a writer, I write descriptions of my characters, and generally draw a concept sketch of each one for personal reference and inspiration for writing, but AI tools have opened up the idea of illustrating far more scenes and locations and characters in the book.)
Good points made here, but I want to say that really this is just the end result of the real problem that's been plaguing users of software for over a decade now: there is no functional consent on the internet. ToS isn't really a choice; every piece of software and every online platform collects data, and there's nothing you can do about it unless you just don't use that app/platform. Often, the ToS isn't even accurate to what the company wants to do, rather they state everything they *could* do, just so they don't have to make you sign a new ToS every time they change their business strategy. It's also hard to justify it as a choice when these companies actively shoot down open source or similar alternatives, either through acquisition or litigation; there is no guaranteed alternative, so there's no real choice here.
Additionally, there is no real safeguard preventing theft on the internet. If you can see the content, a bot can see it too; if you can screencap the page, a bot can too. You can lock it behind a paywall, and that's likely your best bet nowadays, but that limits your reach. As a hobby-ish artist (I also use it for professional purposes but am not a commission or salary artist), I just accepted that my art is in the public domain and can be captured by anyone or anything for any reason and at any time. This makes me apprehensive to post a lot of my stuff, however, especially the stuff I might use in an IP later on. Personally, I'd feel much better if the ToS didn't allow for the platform to share my data without some additional agreement that isn't mandatory for using the platform. Since I can't stop users or bots from collecting it, I'd at least like to have my content and info safe from the platform or software itself. I think we've reached a tipping point with privacy concerns, so as long as people keep up the pressure I could see some beneficial changes coming to platform policies in the future. It's not just the images you post, but the information about you that you do not know they collect (I could have agreed to share my name, for example, but it doesn't mean I consent to you using that to find my email from a database), so it's an overall privacy issue imo.
As for AI, though, I understand both sides. I prefer the open-source approach because it allows the AI to be art in itself after users are able to refine it for a given purpose; stable diffusion I think will go the path of Blender and surpass the paid alternatives. Also, most people who use open-source software already follow unspoken rules about plagiarism because there are no safeguards whatsoever in those programs. For the paid AI like Dall-E and Midjourney, I understand why they need to monetize it; the program is the product of their work, so they should be compensated, and I understand preloading a dataset because it's a core part of the convenience that makes up their competitive edge. You can't just remove all the training data and have users go hunting for thousands of images just to use their product. I understand the artists too; it feels wrong to have your work used to inspire a mass-production capable art machine. I'd also argue that this can extend your reach; personally I would feel flattered that people are trying to emulate my style, and so long as they share my name alongside the AI art, it'd expand my viewership to niches and groups I don't or can't fill with my current portfolio and skills. In the end though, the ethics of AI art are entirely in the hands of its users. Plagiarists will plagiarize, those using it for inspiration will continue to do so, those using it as a tracing sheet for learning will continue to do so. It's the responsibility of the community to hold each other accountable, and I just fear the legal system getting involved because they will inevitably do something stupid like more heavy-handed requirements for image licensing
Art Generators: I make this cool image for my twitch subs of my dog who passed away.
Artist: you hate us and want to enslave us. You want to make us less than you.
I'm glad more artists don't have thay guy's mindset
Is it fair for me to draw then use AI to enhance or change the drawing somewhat in order to quickly draw something from my drawing?
I want to start using this to help me draw or add color, since i've never been able to learn it properly. Is that fair? Using it to help my drawings? Making make some money after all these years?
Probably but the image generators were created using misappropriated / unlicensed assets of other creators. Using software based on misappropriated assets seems unethical to me. The software developers did not even bother trying to make it ethically.
Don't let the people fear-mongering about this technology scare you. The companies who trained the AI did not "misappropriate" anyone's art. If you are not comfortable using someone else's art for reference, then you shouldn't use these technologies, but if you are comfortable with that, then I don't really see the difference.
Whether it's you personally viewing the copyrighted images in order to produce new images, or the AI training on the images to produce new images, I don't really see any difference ethically speaking, so long as the images produced are genuinely unique.
@@mf-- Flat Earther tier logic. Nothing was 'misappropriated' lol, unless you think looking at reference images is also an example of 'misappropriation'
Thank you from the bottom of my heart for this video. People like Steven only cement my misanthropy against society, especially artists like him. We have this problem with his kind in the gaming industry and it is one of the reasons I despise the industry's freelance scene and the so-called industry "professionals" in the AAA scene from the likes of IGDA and it's chapters, Kotaku, Polygon, the jokes that are the AAA companies. They are clowns and often abusive to anyone that does not share their zealous mentality complex. Some of these freaks will try to cancel and destroy our lives for simply not liking our faces, hence their Cancel Culture. This Art Renaissance has honestly rescued the free peoples of the gaming world from these creeps who hijacked and occupied it in the first place in the 2010s (or arguably the late 2000s). I know any actual artists who care about creative freedom and values free enterprise will have nearly no problem with this breakthrough. It is only the elitists and the zealous prejudicial ones that are envious and afraid of this change, because they know they are a major key threat, and it affects them. The evil is shouting it's way to it's grave. It is time for them to pack their stuff and leave us alone, so we can truly thrive as a free world again.
No actual artists will have a problem with AI as they currently already do. Thrive? In a world where AI overtakes real artists?
@@Jdudec367 Get lost with your deluded exaggerations and fear-mongering. No one in their right mind cares!
I'm sorry but I strongly disagree with your counterpoint to AI not replacing artists. I understand that you might think the artist will just be the people using the AI as a tool but it's not designed that way. The AI is developed by computer scientists who couldn't care less whether artists have an upper hand in using this software. They design the software to sell to masses, not artists. So essentially the goal is too make the best images with the least amount of knowledge or experience because that is the optimal product in this line-up. That is what sells...
Nobody wants to pay an artist to fill out a form of text and click "generate". People just want to have beautiful images and the thing AI allows those companies to do is to make creating the images a personal experience. This is something artists can not compete with however you try to spin it. Because art is subjective... you can not order an artwork from an artist and expect a perfect result which matches your exact idea/concept/though. But you can order that from an AI because an AI does not have own subjective preferences. The bigger the training set is the lower its bias will be overall. So as an artist you will compete with your artwork against (essentially) your customers artwork... I don't think, I have to tell you that you'll loose.
People do not “just want to have beautiful images,” they need specific images to a brief. I’m not sure if you work in the field, but if you did you should know that “beautiful images” is just one part of a very large job as a professional artist.
@@done.6191 This was a simplification. What I mean is that most people who would request an artist, they don't really know the exact requirements what they looking for. Even if they know, it is pretty difficult to express it. But in the end whether they like the result or not is subjective.
So you as an artist do not have any edge over artificial algorithms which can generate piece by piece until the customer is fully satisfied. Because this eliminates the loss of information in communication between artist and customer.
It is also technically possible to adjust the algorithm dynamically to any customers individual preferences during usage. That means the amount of work they need to put into filling out forms can reduce over time to get requested results. There's no way any artist using the same tool could achieve that while meeting multiple customers subjective preferences.
The tool can also be marketed as providing any customer an individual experience and you get the advantage that any image or other piece of art produced by it will have a different status. Because the customer will not just have paid for it but they will produced it themselves (in a way). So the subjective value of every piece created by themselves will be higher than anything in contract with an artist using the same tool.
So in the end these algorithms will change the art industry completely, thinning out professional artist as job itself. I don't think it will make physical art by professionals obsolete and some very popular artist will likely stay relevant. But I don't see high chances for artists currently contributing in video games, web design and other low budget contracts. Because using the AI will be more economical and people won't use an artist as operator producing higher cost.
@@TheJackiMonster I have a huge edge over AI, as I can distill extremely complex briefs in a very short time. And AI cannot complete this task, as the time for this to happen: "artificial algorithms which can generate piece by piece until the customer is fully satisfied" approaches infinity.
It isn't that it is HARD for the AI to do, it is that it CANNOT DO IT AT ALL. Until an AI can process language as a human, IE essentially until AI can BE human in all but flesh, it will fail at complex tasks requiring this fuzzy thinking and free associations that humans can do naturally, and do very well with the right practice and training. What AI will do is accelerate the pipeline, as it has been doing for a decade now, by reducing the step-count for mundane tasks. It may be that one day, every artist will become a part of an AI/Human team, with the AI handling grunt work and the Human functioning to put all the pieces together, ensure visual cohesion, produce concepts the AI can't do well, and overall art-direct the bot. Ideally, the bot and human will become like coworkers, and learn from each other, developing a style and technique unique to that specific pairing.
Also, game and film art is probably one of the HIGHEST budget contracts, not the lowest. It requires a constellation of skillsets, all meshing together in a complex technical and artistic environment, to do properly, and as such are generally well paid. True, they don't get paid $90 million for a picture of a dude staring at a pool, but that's because the fine art market is a huge scam/tax avoidance scheme.
The tools will improve, and yes for some tasks "I need a red coffee mug with a pencil in it on the table," they will be automatic. The rest will either be the sole domain of human artists, or will use Human/AI teaming to get the job done faster.
@@done.6191 If you really think you have an edge over AI because you understand human language, you seem to have failed to understand my point. You can completely eliminate the latency and potential information loss of communication by using AI tools. Because a customer does not need to speak with any artist at all to achieve their goal. They can simply tell the machine what to do.
But besides there is actually a lot of research going on for algorithms to process and interpret the human language. So it's not a stopping point for anything in this process either.
However it is far more likely that any human will just learn to provide a machine the necessary pieces of information in understandable terms for the machine. Effectively that's what computer scientists do day by day. So don't tell me: "No, only an artist can do this part!" or "Artists can do this part better!"
Maybe in some cases experience in art will help quite a bit to produce targeted results with those algorithms. But as I stated before: Computer scientists will develop those algorithms. They will implement the user interface and all options to configure the usable process. They will design it in a way you don't need to have any experience in art at all because that is what sells. So no, there is no edge over AI which matters here. Because artists won't have any edge over other human beings without their experience anymore despite the produced results will be art. That is my argument.
I'm not saying the machine will replace you as an artist but it will kick you out of business. Isn't that kind of obvious looking at the descriptions of those tools? Guess why I'm advocating for artists claiming their copyright here?
If you think this discussion is about human vs. machine, you are wrong. It's about human vs. human because artists are certainly a cost in production quite a few people like to get rid of with automation if possible. I mean there is a whole sector of research about procedural textures, photogrammetry and others which partially allow that already.
@@TheJackiMonster “you can eliminate the latency using AI tools.”
“You can recombobulate the fruction with an XG modulation.”
These two statements are essentially equivalent.
There has been ongoing research into human speech recognition for decades. It is still not even close. AI won’t kick me out of business anytime soon, until it kicks EVERYONE out of business.
Just saying stuff doesn’t make the words true. Scientists are working on these algos, as they have for decades, and they’ll continue to try to bottle lightning for as long as people are willing to give them resources to do so. I’ve understood everything you said, I just don’t think you are right. Computer scientists are not artists. As such, they won’t be able to surpass the best human artists (and in some tasks, even the worst ones, ever, until they create synthetic life. At that point, all bets are off.
There is so much hand-waving in your view here that I would mistake it for the Queen.
9:21. With due respect, I have to make a solid objection to your counter. I'm for my fair share of usage of AI elements, let's get that 100% clear, but this argument REEKS and I wanna call it out hardcore.
EULA's are absolutely legally binding, however, to immediately claim that somewhere along the line that all this data becomes EULA compliant is absolutely absurd. One of the biggest platforms, one of the top prompts used in most AI art even, revolves around a website with the following statement in the Terms and Services:
"As between you and Epic, you will retain ownership of all original text, images, videos, messages, comments, ratings, reviews and other original content you provide on or through the Site, including Digital Products and descriptions of your Digital Products and Hard Products (collectively, “Your Content”), and all intellectual property rights in Your Content.".
ArtStation itself has it clearly written in its legal dialogue that all rights are pertained as originally distributed prior to its uploading on the website. If somewhere along that line you might use a piece of software with some sort of EULA clause that might be applicable, say perhaps an Adobe product, it would be through Adobe you'd have to get those rights and NOT through ArtStation. ArtStation has its fair share of traditional media content on it as well, which could easily make the process from paper to ArtStation short enough that the rights would inarguably be only yours.
The REAL reason most of these datasets are "legal" actually comes down to Fair Use, but that's an important distinction to make. Many of these datasets, such as Stable Diffusion's checkpoints, run under GNU GPL licenses, and therefore can be distributed under educational licensing. Stable Diffusion also leaves this little nugget in their license you ought to read "no event and under no legal theory, whether in tort (including negligence), contract, or otherwise, unless required by applicable law (such as deliberate and grossly negligent acts) or agreed to in writing, shall any Contributor be liable to You for damages, including any direct, indirect, special, incidental, or consequential damages of any character arising as a result of this License or out of the use or inability to use the Model and the Complementary Material (including but not limited to damages for loss of goodwill, work stoppage, computer failure or malfunction, or any and all other commercial damages or losses), even if such Contributor has been advised of the possibility of such damages. ".
In lame man's terms, all this license says is "Oh, yeah, you can use it commercially, but if someone comes to sue you or people start hating you for using it we're not responsible."
DALL-E's license says something very similar: "No Infringing or Harmful Use. You must comply with our Content Policy, and you may not use DALL·E in a way that may harm a person or infringe their rights. For example, you may not submit Uploads for which you don’t have the necessary rights, images of people without their consent, or Prompts intended to generate harmful or illegal images. We may delete Prompts and Uploads, or suspend or ban your account for any violations. You may not seek to reverse engineer DALL·E, use DALL·E to attempt to build a competitive product or service, or otherwise infringe our rights. You will indemnify us for your use of DALL·E as outlined in our Terms of Use."
To be completely frank with you, that argument you put up there is from my understanding as someone who's been working all semester on a project discussing the legality of AI art, very, very wrong.
Also, one last little snip at ya if you don't mind too much, while the AI might not carry the images itself with it, there's rules of consent around even allowing an image to be studied for reproduction. Unlike humans, algorithm generations do NOT fall under Fair Use, so the only argument that allows this practice to continue is the same one that got this AI here in the first place: this is a technological study.
If you want a legally solid model, it wouldn't be impossible, but if you can type "Jim Bob ArtStation" into the model and you get a good representation of Jim Bob's work, EVEN IF IT GOT THE WORK FROM ELSEWHERE LIKE GOOGLE, so long as Jim Bob wasn't the one that put it elsewhere, he's well within his rights to mark a line in the sand in most cases. From there the arguments branch out into misleading Terms of Service, the ethics of Google's automatic image-grab algorithm, yadda yadda yadda. The point is that the line isn't as simple as "You signed up for this."
I like these AI, I like technology as a whole and I think it's a fantastic thing as someone who struggles with dysgraphia and fine motor skills in general that there's a place for people like me to generate imagery we can use as inspiration for our own works. That being said, this model doesn't work without those images, even if it doesn't carry them at the end, so at best commercial usage of an unedited AI prompt I'd wager is "legally dark grey".
Love your stuff, dude, and there's no bad blood on my end, I just want to make it clear that this argument ain't it, chief.
I feel like the potential to use the technology irresponsibly isn’t thoroughly addressed in these types of discussions. I’d personally find it concerning if 5-10 years from now AI art could imitate my art and pass it off as something drawn by me. I wouldn’t be able to prove otherwise and the soul and identity of the artist tied to the work would be forfeit. I’m reminded of the story of how not even a week after Kim Jung Gi passed away, his art was fed into an AI model and his style which he spent decades honing, became accessible via text prompt. I think as time goes on AI art as well as human generated art will begin to value showcasing the transparent process of creating the artwork. It’s naive to think that everything we do creatively can’t eventually be replicated, so is the value of the work we currently do subject to change as we become “obsolete”? If certain aspects of the process of making art feels monotonous, if it feels like work and isn’t fun or doesn’t leave you feeling fulfilled to see how you’ve improved as an artist. I don’t mind that you’d rather have a program do it for you. That’s entirely up to the individual. I’d be happy to have AI UV unwrap a model, or skin weights, or hook up shader nodes. But dedicated creators deserve to have their art identity protected and not emulated without express consent. Most tech companies are keen to protect their products from unlicensed usage. I can’t make an art asset in their software and legally sell it without owning a commercial license of the software. It’s a transaction at the end of the day. In the future, I hope artists can leverage their products in the same way before being stored in an AI database. As long as AI is done responsibly, I’m fine with it.
Beautiful comment, Austin. I've also heard the point made about artists providing footage of themselves creating the artwork to prove it's actually human-made. This would solve a lot of problems around authenticity. The question down the line is...will clients and fans pay for authentic artist created artwork when AI can create nearly identical art basically for free?
I see a bleak future when artists can only make money by 1) selling art to their already established fans, or 2) targeting high price buyers who want that most novel and chic collectible...the human made artwork.
If you look at internet-age revolutions like file sharing (Torrents etc), digital music downloads, pirating...you won't see much integrity. I'm afraid AI art will be the same. Fiction, Art, Music and eventually Animation will be AI opensource free-for-alls. Artists who devote their lives and talents will be reduced to referencing data points for AI prompt jockies who profit by cheaply and quickly making endless iterations which can match any art style/content.
There's no way for artists to opt out of being mentioned in AI prompts that I know of. I read that there was a move to make this happen by a group of concerned artists but even if the AI opensource companies agreed to this, who actually thinks AI users won't get around that?
First of all if you shared your art online then it's already out there, people won't need an AI to copy your style, just some technical painting skills to make a reproduction. And if someone claims their art as if it was made by you and try to monetize it off your brand it's a crime no matter if an AI was used to produce it (and i don't count artist's name used in prompt because no one would pay for art because someone's name was used as a prompt only if marketed to be made by certain artist or because of the nature of final work, which is only partly inspired by a certain name in a prompt, no more than any of your original art was inspired by every art piece you've ever saw). So it's really not an argument against AI. If someone does it you can persecute it however you want, leave a disclaimer that your authentic art can only be found on your distributing places (site, social media or etc.) and anyone selling it otherwise is scam. If you wanna leverage in court that your "artistic style" is the same tangible concept as a source code of proprietary software and anyone using it without your concent is violating your copyright, then good luck to you and an art community as a whole. As music industry copyright bullshit clearly showed already, you won't be able to post shit anymore because you'll inevitably infringe on someone's style sooner or later and a swarm of copyright trolls will take any money you'll try to make using your art. It might be disheartening that you had to spend a lot of time and sweat to learn to do things that people can just accomplish with a "text prompt", but people care about it as much as any other automation field, so you have three options: be a luddite and reject any technological progress at all, be a hypocrite and care only about machines replacing exactly you or be appreciative of wonderful possibilities that new technology will present us and embrace change by changing yourself. Eventually people will have to fall in either of these categories so you must decide for yourself on which side you have more to gain for. There will be communities and marketplace for AI rejectors for sure but will you be successful in those is a big question. Rising against AI might sound like great idea but it will ultimately fail as most of luddite movements before, and leave "purely human art" as a very niche market share and it will be hard to compete outside of it if you reject it long enough. Also who says that there won't be an AI that will be able to fake "creative process steps" for every art piece you present it. Even before Dall-e/midjourney/SD boom there was already AI algorhythm that performed "neural style transfer" generating colored simple shapes on top of each other just like oil painter would do starting with big strokes and layering more and more small ones to add details.
Wait. You think AI would be able to copy your style in 5 Years?
🤔 Emm... thats pretty much possible already.
@@cesar4729 I meant in the context of a hypothetical impersonation of a not so famous artist. Like an AI art thief account. But yeah that’s probably already a thing at the rate tech is going.
Start cryptographically signing and watermarking your images; pick a good enough algorithm, and the world would be in much bigger trouble if someone figures out how to break the cryptography than just having some perfect forgeries going around, so if it somehow happens, proving art authorship will be the least of your worries.
The thing is - if AI makes art-makers obsolete, the people who will be making money will still be the same sort of large companies who once employed artists. They'll still make money selling entertainment, etc... but it will be individual artists who can no longer be independent (and that's hard enough as it is). That's a legitimate concern...
On the other hand, if it is the large corporations that become obsolete (say.. it becomes feasible for average Joe to make a beautiful 3D game on the level with AAA titles), then maybe it's a different story.
But we're still more at a place where it seems like proprietary companies will benefit the most. Force 'em to go open source, take away their 'right' to use art not in the public domain, and things become much more humanistic and far less misanthropic.
without "evil big companies" who will pay artists for work? Who will provide resources?
@@sychuan3729 Who needs 'em? I could live without Fortnite and Mindcraft (kids would certainly benefit) and wouldn't mind a few more indie games and comics by small independent teams. The worth these 'evil big companies' offer comes at the expense of culture and value, as they skim the bottom for the most addictive, most gruesome (horror movies today...), most lurid, most vile.... yeah I have no value for these big companies or big Hollywood.
That's why people should support Stable Diffusion over Midjourney, or Dall-E.
SD is open source and being open source has already has proven to be a superior strategy.
Blender, for example, is already way, way more popular then the pair alternatives, mainly because it's entry barrier is next to zero when compared to the alternatives, but also because they added new cool things (like geometry nodes), but also allow for easy extendability.
Overall always support open source over proprietary software, because this way the software is for everyone by everyone.
@@graythistle wasn't this one of the few points the two guys in the video agreed upon? An independent illustrator that has passion for painting grotesque realistic horror (what AI does best apparently) is not gonna suddenly have passion for composing comics or designing games for you to enjoy.
@@nocturne6320 I agree and think that the solution is Stable Diffusion's Open source model, which makes it a more reasonable proposition - the problem is that anyone can cut corners and replace an artist who has spent years cultivating a style that can now be ripped by AI. That is at best misanthropy. In all fields we should be protecting people from AI. This is what separates the Postmodernism hatred for humanity with a brighter future that will benefit all.
As an artist, I am walking a fine line if I use the art of a living artist as a basis. I could easily copy a popular artist like Loish if I wanted to, but there would be a lot of backlash if I tried to make a living doing that - and rightly so. People would be telling me right and left that I'm a copycat. I walk no such line with the public domain and can mine it for whatever I want as I like - the AI and the person behind it should have to do the same, and suffer the same backlash I would if they don't. In fact, it should go further: that backlash should be legal and if AI produces a bunch of copycat Loish art, she should be enabled to sue the offending company into oblivion.
No one really mentions the physical material (severs) and energy costs (electricity) of running massive AI art farms. The more complex some thing is the greater the expenditure of resources is necessary to make it. So that silly 'mega feed' Steve mentions is kind of unfeasible, generating an entire movie off a conversation isn't going to happen it would require FAR to much power and render time.
Love how visual artist hope to be regulated like music artist.
They just have to try what is considered "original" in the music industry, and then they will quickly change job because they will not possibly using any copyright material as reference or inspiration
The original video against AI felt like it was 50% hyperbolic wrong information and 50% condescending elitism mixed with fear mongering and alarmist propaganda.
I stop taking people's arguments seriously when they start using massively manipulative emotional tactics, and most of what I could see being said against AI after that, is misinformed bollocks.
I've had people steal my art before, both physically in person and online where they claimed mine as their own creation. I've had people sample my art for use in their own compositions and photoshopped collages. AI art isn't even on the same continent as any of those behaviours. AI really is no different than doing studies of someone's art until you can replicate their style, just as actual artists do.
I was already for AI art, though I was leaning towards the "should be opt-in" side of things. You made a really good point many times over about how we already opt in when we click "I agree" to the ToS. I have long accepted that for things like my personal data, but I never considered that extending to things like uploaded artwork.
@@Real-HumanBeing Exactly.
Your uploaded artwork *is* personal data. That being said, it's only very small communities that actually try protecting their data, and I don't really expect that to change.
So, if some future AI technology scrapes the internet of everyone’s personal data and allows any random person on the internet to do what they like with it, you’ll be ok with it because you clicked ok to some TOS years before the technology even existed?
Thank you for taking the time to get this out there. I am an artist and a writer and I love having the AI tool to help me realize my work.
I think you're missing a lot here. This is a long comment, but please read it.
9:10 That comparison doesn't work, as LAION relies on images taken from the entire internet, and to my knowledge, the Google ToS doesn't include "Will be used to make an AI that can knock out images under your style". LAION takes most images from sites like pinterest and etsy.
11:30 There are only so many eyeballs for so much content. No one is going to care about someone's 300 pieces a day, that's not how attention works. The end result will be that FAR fewer people will be able to live off of doing creative media, with handful dominating everything. 1.4% living off of their art (will get to that number later, it ignores photographers, designers, and other soon automated fields)? Try 0.015%. That's the number of twitch streamers that earn as much as the average US household, or more. Given how streaming creates the most content hours for least effort, this is the best comparison, except that knocking out AI images and stories will be even quicker for more content. A future where basically NO ONE can live off of their passion.
14:30 The main gripe is how there is no protection for artists, like Greg Rutkowski. The "artists don't make that much of the dataset" argument only works if you ignore what the AIs to 95% are used for, and how artist's names are directly being used.
27:30 "Not using it is career suicide" -- So you're admitting that the vast majority of artists, designers, photographers etc will be put out of their jobs? It is easy for you to be so nonchalant about this, but don't wonder when Steven isn't sharing your ungrudging attitude.
28:50 His main agrument is about it taking images from people that never agreed to their usage in this way. You have yet to produce the ToS clause that allows this, and given how LAION is the base for these AIs, it's missing the point.
31:30 Of course they're not going to say that they're about replacing artists, though that is what it comes down to. The farmer analogy is disingenious and doesn't work.
31:45 "artistic people will be even better" -- I don't see it. It already pools from the greatest artistic minds, making far more decisions than are put in, and better ones than most could come up with.
35:00 He never makes this argument against automation, but I will make one. The more we have of something, the less it is valued. Why is it that during the age of instant access to great masterworks in beautiful colors, instant photography of gorgeous landscapes, we live in grey, uncared for homes, drive grey cars and use grey appliances?
41:10 So you envision a future where basically everything is done by program, and that's a good thing? Is it so weird for you to consider that there are countless people out there that love doing it themselves, and that have worked their entire life to get a job in this? When I see an art studio of hundreds of people that can support themselves doing something they love being replaced by one "quality assurance" guy occasionally checking in on the outputs of a machine that basically does everything itself, having more knowledge of current trends than any human could, that is simply dystopian to me. How far do you want to be removed from making something yourself?
1:00:00 "Feelings" -- Your side also hinges on feelings, because what is making stories or consuming them without feelings? I don't know why the idea, that some people love spending the time on drawing their own comics, putting in all their effort and care and doing every single stroke seems so alien. To me, the idea of letting a program do that is just shallow. If you really love your creation, wouldn't you want to make every single detail yourself? Your own argument is about how important the feelings of neverending consumption of AI content are.
1:04:20 Since diffusion models are prone to overfitting and memorization, you'll find bloodborne generations that look almost 1 to 1 like bloodborne cover art.
1:06:30 "Human theft is worse" -- that is simply not true. The most important clause of fair use is the 4th, about how cited work affects the value and market of the original. How is the value and market of Greg Rutkowski being affected by "cited work" being used to make unending copies of his artistic identity, to the point that he has stated that he worries about his employment? Greg is not going to care that you copied his image onto your desktop, he may even appreciate it. He may also appreciate a person making a work in his style as a tribute. Even if someone 1 to 1 used his work to get employed, he would barely be affected. That is "human theft", and it is completely benign in comparison to what AI "made by Greg Rutkowski" prompting by the millions does to the value of his work.
1:15:00 I highly doubt that they would've used copyrighted images, let alone made a program where you can enter an artist's name directly, if the artist industry was as capable of suing as the music industry. Do you really think the art equivalent of a Taylor Swift AI would EVER fly?
1:19:00 "Not a real artist" -- The crux of this argument for me is how minimal the human input is for (almost all) AI images. When you're actually drawing something, you inevitably have to make thousands of tiny decisions, that now AI makes. I would go as far as to argue that the idea is less important than the execution. Is a LEGO movie a good idea? Or a guy fighting crime in a batsuit?
Not by itself, it's the thousands of small creative decisions that turn it into a worthwile endresult. I don't think AI prompts can't be art per se, but they almost never are. When you pool from the greatest artists that have ever existed,
the AI is bound to make decisions that far exceed those of the average prompter in skill and uniqueness. Most prompting is only commissioned consumption rather than creation.
1:25:00 Please give me the ToS part where it states that images uploaded to pinterest or etsy will be used to make AI models that will make images themselves. Even for google cloud/photos, I don't think it exists, and again, LAION doesn't rely on google cloud, and even Google's imagen relies on Microsoft's COCO dataset and LAION.
1:30:00 "There is nothing stopping you from making art however way you want to make it" -- When I have to spend most of my life working a job that has nothing to do with art, for most of my life I'll be stopped from making art. You may say "That's 1,4%!", which excludes designers (fashion, covers, products etc), which excludes people with part-time jobs and photographers among many others, and which doesn't consider that the same thing is coming down the pipe for voice actors, musicians eventually too, writers, sculptors, actors, animators, and every other job that people actually love to do.
1:32:00 "If you don't make your own colors etc not a real artist" -- Refer to the earlier point about the majority of actual creative decisions being replaced. This comparison doesn't work.
Don't be mistaken. This is NOT like any of the earlier cases. This is taking someone's work to put that person out of their job. If you don't find that inherently worse than coming up with a novel invention like the car, I don't think this debate will go anywhere. There is no historical equivalent to this.
the thing that makes me most optimistic is how unfullfilling prompting is. sure now it's a novelty, but what about a million generations down the line? I actually thought before this all came out at this level, taht it would be like a release valve with hundreds of millions of people instantly going for it, but i think most understand it's a hollow victory
@@captainobvious9902 Huh, yeah, like inflation of the dollar...
Thank you for this comment, I read it all. I haven't watched the video. I did watch Zapata's last week or so, I liked his the best (I'm a hobby artist). It's not "fair use" if it takes a person's livelihood to train on, and then puts that person out of work.
I have to wonder why Disney isn't putting their foot down, I saw a lot of Disney stuff...
@@jennyheidewald5006 thanks, appreciate it
@@michaeldufresne918 Yeah, I feel humans are constantly chasing after that high. The next cool new thing.
Hey Royal Skies. I hope you respond to this, though I realize that you are a busy man. So I've been looking and talking to a lot of people in the AI forums (people on both sides of the fence) on reddit and stuff, trying to get a broader perspective on the subject. I've really been trying to put my mind at ease about all of this as the only real hope I hold onto in my life being able to find a profession creating art. As you are the one to introduce me to AI art, I was excided to what this video as I find myself having a lot of the same thoughts and feelings as the guy in the other video, and truth be told, after hearing your replies, I feel incredibly worse, and I think the big disconnect here is perspective. And please understand, I don't think you're perspective is wrong by any stretch of the imagination, but its missing a few angles, as is the other guy.
I think where your argument falters is somewhat in combination of the feelings (and please hear me out) and gatekeeping. While I 100% believe that art and the ability to create should always be available to everyone ever, I feel that a gate is extremely important too and I'll use your examples to explain why. Going back to the Fire Fighter, the reason that they shouldn't be able to create great art so easily at such high levels is because they chose the path that doesn't/didn't allow it instead opting for another path with its own rewards. That Fire Fighter, like all Fire Fighters is an absolute badass! A Hero! And we/society all know that and acknowledge them for it. They can lay their head down at night knowing they made a difference. They can do that because they chose to spend their time becoming great at that skill set. Artists can't run into burning buildings again and again and be those heroes because we didn't train for that. Its a gate we didn't pass through, and that gate keeps unexperienced people safe from the flames. Artists chose to spend time acquiring art skills instead, and whether people want to admit it or not, I'd imagine all artists want as many people as possible to see their work and praise them for it. It makes us special. The Fire Fighter can raise their head high for what they do, and we can raise ours for what we do. And I think this goes especially for artists as the majority, at least from what I've seen in art school, forums, ect, grew up as the "not cool" kids in school. This is supposed to be what makes us special, what helps us shine, but AI Art is like a super power and in the words of Syndrome from the Incredibles, "Everyone can be super! And when everyone's super, no one will be."
The other major part I feel like you really, really glossed over is the fun factor of art, but I think I get it. You are an end product kinda guy. That's cool. Nothing wrong with that. You're a badass, a successful badass (so you're lightyears ahead of me). You be you, but you aren't seaming to realize that not all of us are. You ended the video saying that artists will always be here. That we'll always see that kid doodling in class, and you are 100% right, but that's because grade school is boring and drawing isn't. The dream (and I do understand how much work being in the 1.5% of profession creators is, I'm not low balling it here) is to get to be paid to do that. To walk into your studio and just create for hours on end, to be paid to basically have fun (you know, that saying about if you love what you do you'll never work a day in your life) and for some of us, creating is the process as much as it is the end result. I'd love to just get paid to sit and model a dragon for hours on end, but if the AI does the fun stuff for us....how is it any different than any other job? Hell, you even said that prompt crafters will probably just end up as AI managers down the line, and maybe I'm alone on this, but manager style duties is almost the opposite of being an artist. One of the big selling points of artists isn't just that they make great art, but that they could make it fast. It gave them leverage for pay. That's gone now. So what's going to stop companies from cutting artists down to a skeleton crew and lowering pay? Once those doodling kids in class see that drawing can't make them a living, well they're probably just gonna chose another field, especially if they can just toy around with AI.
I dunno. Maybe I'm way off base here, that's just how I personally feel right now and really needed to get it off my chest, and while there's more I could say, it'd be mostly personal feelings and short sightedness. Thanks to anyone who read this far with an open mind. As a lousy artist who can really only benefit from the AI (and I do see all the cool ways to use it), I really, reaaaaaly hope AI turns out to be nothing but greatness.
This. I agree with everything you said here. strangely enough I was reminded of that exact quote from the Incredibles. You can tell when someone put their blood, sweat, and tears into something they are passionate about. I hate the fact that i spent so much time learning how to model, texture, wight paint, UV, and several other skills just for an AI come in and do it all in a matter of seconds. Its extremely frustrating and makes me wonder if there is really even a reason to continue doing what I love. Maybe I'm being dramatic, who knows.
@@jihyolove9638 Yeah its rough. I don't know the answer.
@@sterlingbirks9101 Yeah, when those college bills come in and there's no job in sight, it can be rough. I'm still paying off my loans 10 years later, though I made lots of mistakes so I'm not a good example. Still, I feel you. Royal Skies is super cool and super talented, but doesn't seem to understand how this threatens people who aren't as talented as him. If it helps, I've see work flows that use AI art in ways that still require effort and skill so I'd say look those up so you can still get some satisfaction from your efforts.
If you enjoy the act of making art, nothing will stop you from continuing, in the same way that some people still make their own clothes, or blow their own glass, or whatever. And some will pay a LOT for that, despite automated techniques that exist to produce both.
@@sterlingbirks9101 The worry is mostly unfounded. "Specialized real world art" is ALREADY the only way artists make a living, regardless of genre (visual, audio, performance, etc). No one, except Redbubble cell phone case folks etc, has a huge market for "pretty pictures." They need something very specific, to fit into a workflow of often dozens of others, and the AI is no where close to closing that huge loop on its own. Source: I've made 100% of my wage doing art professionally, for 25 years. People who focus on craft over concept rarely go far, because no one wants boring art that doesn't hit the brief, even if it is well-executed.
I know this video is trying to keep the ethics confined to AI art in a vacuum, but the "Consent to EULA" argument is an appeal to Status Quo argument. The reason people skip past TOS agreements is because all those agreements contain more or less the same stipulations, and the only alternative to agreeing is to not participate in modern society.
This also runs counter to the ethical underpinnings of intellectual property, which was initially conceived to stop people from just stealing newly created works.
Yeah that argument was a bit weird to me as well. Many of the websites you upload your art to don't have anything to do with those companies making these programs. So I doubt those terms of services include being ok with ones art being included in those training data sets. Who knows how that's gonna play out though and if using someone else's pictures is considered "stealing"
@@discipleofschaub4792 Like, would he make the same argument to Eve6? A 90s band that sold the distribution rights to their songs before Spotify made those rights worth more than the CD royalties?
"But if you didn't want your work to be seen by anyone and anything you shouldn't have make it public." Thoughts like this is why AI ethics is a thing.
@@maxicornejo9675 AI ethics is stupid and just a way to preach certain ethical positions when they only signed up for the knowledge part. I'd make the case that if I had read Harry Potter, that likely, in some way, affected my ability to write english, and that, in some way also affected my writing style. In essence, ai isn't stealing any more than I am.
@@Cecilia-ky3uw I have three questions for you. Do you think merit is important? Do you think that an artwork provides more value to an algorithm than to a human? If you do, where do you think that value comes from?
I think the counterpoint from you against stolen art is pretty much a straw man. He never said anything about hacking or uninformed platform usage. I agree with you that people give out consent how their own work can be used way too easily (let's say they post it on Instagram for example). But the issue is still valid. Those neural networks are trained on art without consent. You can actually look up the database used for stable diffusion because their set of tags and links is open to use. However when you look up the actual images behind those links, they are not owned by the developers or researchers. Those are public images in the internet (public as in accessible) but there's no information about the license to be found. That means legally in quite a few countries, you should expect them to be copyrighted and therefore they should not be used for training of AI without explicit permission.
Otherwise every artist on the internet would need to build up private paywalls and start suing other people like crazy to counter that. It's a massive copyright violation and it needs to be addressed. I'm very sure of this because Github copilot from Microsoft did the same thing. They just trained their neural network on public source code and sell the tool now. However it is impossible for them to always check licensing for each piece of code. So it is extremely likely they violate licensing by selling generated source code derived from code which potentially explicitly denies that by license.
Then if a company like Microsoft doesn't care about this, why would any other company request permissions for artworks here? You don't need to hack anything. If you just crawl all the images from DeviantArt to train a neural network, you still violate copyright. It's that simple... would be a totally different story if they had only used images from Pixarbay or similar.
BTW in copilot's about page it was mentioned that the system "shouldn't" generate copies of existing code or public personal data from training material.... "Copilot was trained on publicly available code..." and some dancing around the issue of ownership... "GitHub does not own the suggestions GitHub Copilot generates. The code you write with GitHub Copilot's help belongs to you, and you are responsible for it."
No it's not. Training using existing art does not violate copyright. The same how sampling a popular song and modifying it doesn't violate copyright in the music industry. There have been court cases regarding sampling, and more often than not it gets overruled as a moot point.
The AI may use your art to train, but it is not giving out the art to others with a copy and paste, nor is it using 'commercially' because the output is always modified. There is no copyright over art styles, which is also a popular point. If someone makes an AI piece of art that looks and feels like your art, its till not yours.
@@devnull_ Of course it doesn't generate existing copies. Still it generates portions being copied. There has been occurrences where authentication tokens were suggested by copilot which shouldn't have been publicly on Github anyway. But that's not the point.
AI algorithms are pretty much always trained without consent of actual owners. Because it wouldn't be feasible to do so. The training sets are huge and asking every owner of data for consent would take years. So this is either done by terms of service or illegally.
To me it's a whole reason not to use Github if possible because I work on free and open-source software on the assumption the software stays free to use. However if copilot gets trained by such code, Microsoft gets paid for it and the result ends up in proprietary software without publishing its code (even though portions might belong to me and are licensed under GPL)... well then AI is effectively just a tool to make license violations legal, isn't it?
So from my point of view none of the algorithms are problematic. It's totally fine to use AI for automatic generation of code, images... whatever. But the training should not hurt the original owners of the data. However currently it does.
I personally would love to see open software which allows artists to train a network with their own images to help them reduce time doing their job. But that's not what is marketed here...
@@DankaDoctor185 It's not safe by copyright either, though. The part that people seem to always miss in these conversations is the word "USE". It's not just about the image itself, it's about the usage of the image. Sure, the AI might not carry around your photo or drawing verbatim, but if it knows enough about you to recognize your name and make a doodle that looks like something you did, that's usage of your work. Sampling and modification are also only fair use under HUMAN rulings, you cannot apply the same logic to an algorithm, there's no heavy legal precendent.
Lastly, since your case is about fair use, I feel the need to let you know the following is a factor in court decisions of fair use:
"The effect of the use upon the potential market for or value of the copyrighted work".
Now understanding what use means in the legal sense, suddenly this is not as clear as you make it out to be.
I'm not going to claim one way or another if it's truly legal, but I will say this point would probably not hold any water.
@@TheJackiMonster Because I like keeping things neutral, I do think it's important to state that the AI does not actually jigsaw existing images together. It kind of works in a way similar to how your brain constructs things. th-cam.com/video/1CIpzeNxIhU/w-d-xo.html
This video is a really good explanation, but the argument that it's actually using other images and reassembling them in portions is incorrect.
It's a really fascinating technology, and while I'm against its commercialization at the current state, a model trained on Pixabay or another CC0 database could be incredible if the steps were taken to filter out copyrighted material.
I have to continue watching your video tomorrow. It is far to great to watch it at 4:30 am on new years day. But I wanted to write something about myself and why I am glad about AI being as great a tool as it has become until now. For me, AI-Art-Systems are plain hope. I am severely disabled due to a multiple sclerosis, have two partially paralyzed hands, cramps and spasms and worst of all: fatique. I had to give up my dream job as an art director in a small advertising company - in fact, I am unable to ever work again. I had to retire and can‘t really get out of my bed and even less my home, without help. I still make art, it takes a huge ass time to make art, but I need to express myself creativly. i wouldn’t have any wish to stay alive with all the chronical pain and being unable to live on my own, if I couldn‘t make art. The problem is, that I won‘t be able to use my hands indefinitely. In the moment, I can skip a lot of time, when using stable diffusion as a tool. When working on the right wording to get exactly the picture, I would paint or create myself and then complete it with my own paintings. In the last few month I more and more developed my own workflow with it and it is amazing. If I would still work as an art director, these systems would make my work life so much easier, giving me time to create even better works for my clients, although I also think, that many companies will employ less artists and focus more on the directing creative people, rather than the graphical artists themselves. There will be new jobs and new opportunities for artists in the future, but yes - some professional artists in the industry will loose their jobs due to better and faster solutions. That is sad, but it happened in every branche, with every job so far over the time. Artists used to be the lucky few, who would never be replaced by a computer, because they cannot create art… oh well, that is now a thing of the past - but this doesn‘t make our challenges special, it finally makes artists like everyone else out there.
To come to a conclusion for today: Steven sounds like he speaks for every artist in the world but only speaks for himself and those not understanding the matter. Either out of fear or with vicious calculation, he shows a total lack of understanding with a pretentious and gatekeeping attitude. And he (in my opinion) reached his goal. Finally everybody talks about him. Sadly nobody about his great drawing in the video and everybody about his lies in one form or the other.
oh well…
I generally agree with you point view, however it seems like there will be an escalation of a big purely digital problem which already exists now - informational entropy. Even now it’s extremely hard work to find a useful information for self education, you have to scroll and sort a tons of bits in order to find a truly quality inside shared by someone. And with all surrounding ai-technologies the amount of produced information will grow exponentially faster than now. There will be no place to search, just pick the best solution proposed by the algorithms trained on your data.
I think this is a huge problem, not in a personal level (because personally it’s amazing to perceive a purely individual things you love), but it will apply to the commercial industry of all sorts and cultural mess. Maybe chip-implants is a solution, but not a certain one. I hope people will find the ways to spend more time off-line with each other, but for now, it seems like atomization to reality is an up-going trend.
Good point. Which I'm sure one of the big software companies are working on already. Ai art already is trained on good vs bad quality so I'm sure we aren't far from the idea of search engines filtering out the most accurate data we are looking for from trillions of data.
I already notice that (such as in online shopping). I think there will be a number of solutions including going more personal and private vs. just cruising the internet and social media.
Love this, subscribing. Btw i like the little radio in effect when you chime in
that effect is from the videogame starfox
The first thing I don't like about AI that you don't have much fine control about the creation as it is now. That's why Zapata is right when he says that it isn't a tool. It is only a tool in a very broad sense, but it doesn't stimulate us creatively and mentally the same as when we are way more involved in the process and in control of the microsteps. So it hinders self expression that way and spreads mental laziness.
My second problem is that AI will lead to a massive oversaturation of the art market (and possibly movies, music, video games etc in the further future) so it will become increasingly difficult to stand out.
You are right in that the truly great artists could benefit from that and stand out even more but that depends on how good the AI will become. Should it reach a certain level of perfection or would even surpass human creation then together with the oversaturation of art everywhere it could lessen the general demand of and interest in art to such a degree that the spotlight artists often relished until now could dim or go out completely and I don't find that motivating for artists, who want to get attention and popularity.
You say that art matters even when you are the only person in the world (maybe together with a few friends) who likes it. And while that's true, it seems a bit involuntarily cynical to me because it's also extrinsic motivation that drives everything we do and we also need external validation to be complete in our happiness. And the prospect of a world where all your artistic efforts are in no real demand anymore is just disheartening and hopeless.
"we also need external validation to be complete in our happiness." I'm not sure this is true.
@@dibbidydoo4318 It's the same reason why having a (good) job gives meaning in life and having none often leads to depression: Because you make or do something that is of value not only to you but also to others. Humans are social beings after all that don't live only for themselves.
@@dibbidydoo4318 You and Royal Skies does not realize it, but just because you don't think it makes you happy does not mean it does not make you happy. When you make something, put a lot of work in it, you are probably doing it to share it with others, this goes beyond painting and such, even just the idea itself. If all the people you will be able to show your idea are just your family, then you will not want to spend any time on those things. You will have creativity that you cannot release because back in your mind it will not matter. A child seeking validation from their parents is normal, but not when you are older. People need validation from others, not just in art, but in most parts of their life, but especially when creating something.
@@lexiferenczy9695 I agree. There seems to be an idea going around that working to make other people happy is a kind of narccissistic attention whoring, but working only to make yourself happy is enlightened selflessness. It seems obvious to me that these beliefs are cognitive dissonance.
With inpainting you absolutely have total control over your work.
1:27 I completely agree with this definition of Art, that is the definition I use after all.
Could you point me to the section of the Icloud user agreement that gives third parties the right to trawl private cloud data? I can't seem to find it... This isnt how the AI data sets is created by the way but I'm fascinated by the multiple directions you're able to wrong at the same time.
Agreed. No idea how he keeps going on and on about "you agreed to the ToS" when there's absolutely nothing in the ToS that would permit it, and as an "AI artist" he really should know better about where the data comes from
Man if only the world was at a point where it can openly embrace this without having to worry about having a home and food, because that's what it boils down to in the end isn't it? You put years into learning your craft so that you can eventually live off of it, all the while AI slowly but surely is automating jobs away and changing fields isn't simple or easy, and we still don't have housing for everyone or food, plus it still is pretty unclear, uncertain, where we are headed with automation. Whether we'll actually be free to do as we want or everyone will be cast into poverty as they're no longer needed.
100%.
Automation wouldnt be so terrifying if it wasn't eroding people's jobs, and arguably at a faster pace than new industries are rising up to replace them. We don't live in a post-scarcity society. The resources needed to feed and house and clothe and care for everyone on the planet exist, but not everyone has access to them, and they are making what you need to live harder and harder to access every year inflation+cost-of-housing+cost-of-food+cost-of-utilities(because inflation includes none of those things) grows faster than wages (which is every year in aggregate, and most years for most of the population)
This is why humanity is going to HAVE to get past this "you only deserve to live if you're slaving away 40 hours a week" thing, and implement something like a UBI.
We're gonna need to stop demonizing "lazy people" because EVERYBODY is going to become obsolete, and it will be happening faster and faster.
People in the US rant and rave about how much they don't want communism or socialism, but that is exactly the kind of things that automating everything is going to force us to embrace!
You know what is funny about that argument? Is when you look at history. People complained when machines replaced factory workers. However, before machines replaced factory workers most people were never given the chance to aspire to be anything more than that level of worker. Only the very wealthy could afford to be artists and inventors. Poor people stayed poor, they weren't offered a very good education because why would a factory worker need a good education?
However, at the time people didn't know that machines taking their soul destroying jobs would benefit them and us, future generations. Imagine if they had really fought against it to keep people in these kinds of jobs. Our generation would still be there, and only the wealthy would have the luxury to be creatives.
My great Grandfather was a coal miner, so were all his children, apart from my grandad who was born as they banned child labour and instead children of all classes were given an education. People complained about banning child labour too and sending them to school. Before people had children so they could send them out to work, children were made to start working at a very young age. It was only possible to offer all children an education because machinery were starting to be invented, that mean they didn't need everyone working.
If anything these programs are being used by people that can't afford lots of art materials and equiptment, its not going to prevent anyone from creating art if they want to and its not going to stop people buying art made by people, if they like it. All it is going to do is give more people access to creating art. Exploring you imagination shouldn't be limited to only those that can afford it.
I don't know what you mean by no longer needed. Art is always trending so what if your art style isn't currently in fashion. That shouldn't stop you from being a creative. Humans have been creating art since cavemen painted pictures on walls, it is part of who we are as a species. Its not going to stop just because we have more tools to be creative with. Even though most people now have access to cameras on their mobile, it hasn't prevented professional photographers from getting work. I know a lot of professional photographers that once worried the smart phones would take their jobs. It hasn't. But it has allowed everyone to explore photography and discover if they have a talent or passion for it without needing to invest in expensive equipment. When the camera was invented lots of painters freaked out that they would lose their jobs, thinking why would anyone hire a painter if they could just take a photograph. Yet painters are still around, if anything there are far more painters now than ever before in history.
@@codeXenigma I get what you mean and the automation of past industries is a fair point, I frankly do not know what to expect so I'm just wait and see, and learning a bit about this AI since Royal may very well be right in not using it being "career suicide". As for no longer needed, I mean that this may cut down several jobs (I mainly see talks of concept artists in regards to this) but of course not only in this field but many, yes it is true that as one thing gets automated and those jobs are lost, other have come up, but it remains to be seen how AI will play out, I do fear Steven may be right that this is not a tool but a replacement, sure you can keep doing it as a hobby and AI does open more creative opportunities to everyone, but if you can no longer sustain yourself because your field has been automated, you are indeed no longer needed, so change fields, and hope that also isn't automated away? And hopefully still have time to be creative.
@@codeXenigma you lose me as soon as you compare companies ripping off artists all across the internet, to child labor...
Art is one of the very few jobs that people WANT to do themselves. You're really gonna compare 30 years of passionate training for something you love, to being forced to work in coalmines?!?
Apples to oranges dude.
You think ai art is no different than when calculators came out? Well calculators didn't enable people to go around pretending to be Einstein himself. There is a WORLD of difference.
1:25:30 I want to be charitable and say that I think you misunderstood the entire point here. If I say "I didn't agree to have my work put into this AI databse, it's sus that they never asked me to agree to putting my work in the database, and I know they will not ask nor allow me to opt out in the future, therefore this bad and doesn't care about me as an artist nor my consent" you responding with "just be happy that they gave you this opportunity in the first place," is equally immoral if I take you at your word and that is in fact how you view this. If someone doesn't want their work utilized in an AI database, that is their decision and it should be respected, whether it "grows their audience" or not.
Do you have to explicitly consent for another artist to view your work that is publicly displayed and then use it as a reference to produce a completely different piece of art though? No, you don't. The consent is already implied.
Really appreciated this upload.. its right on the money. The Laion training model being used is an unethical montrosity designed to smuggle 5 billion pieces of human made art in such a way as to specifically make litigation very difficult. The end result is nevertheless no less damaging to peoples livelihoods. To quote S.Gordon's film, From Beyond.. " "that machine has got to be destroyed"
I see art moving in the direction of youtube, etsy, patreon, twitch etc. More artists will have to become internet personalities/commission artists. And work with ever day people directly instead of working for large companies or studios.
I don't think the mid or higher level commission/entertainment artists will be very effected by this. Commissioners often buy the art they do because of the brand or person behind it. For example, look at closed species communities. People pay just for the "right" to draw a closed species character themselves(not even art! just the right to draw it themselves), and sometimes hundreds of dollars for premade designs just because of the brand behind it. It makes them part of something. Its a badge of honor, a bragging right. It makes people happy. AI art can't hold that level of prestige. If anything prompt engineers are going to receive the same kind of backlash digital artists used to get... Constantly being told their work is "not real art." And having its potential value be scoffed off by the mainstream.
I do think beginner commission artists will be effected tho. I used to sell my work for next to nothing ($5-$30 range) and the clients I got back then where often a LOT more unpleasant than the ones I have now. (some even trying to scam me into working for free). Looking back, they obviously bought for the bargain , not for my style or quality. I can see those types of people using AI for free work instead. (Gone may be the days where artists are asked constantly for "free work"). But that does leave a smaller client base for newer artists. They may have to work longer and harder before they can sell their work at a reasonable price.
I don't think people will suddenly stop supporting artists and all swap to AI. If anything it just gets more people into art and makes it more accessible to more people. I wont sugar coat it tho, a lot of artists are going to have to change the way they do things. The quality expectations are going to rise and Ai brings a whole new level of potential thievery/con artistry into the mix.
I do find the click bait articles exhausting, and the copyright issues give me a headache. The companies leading the way make me nervous as well. But I don't think we should all just give up or spend hours on end arguing over "what is art." (not that anyone here is doing that, I just see it A LOT). If anything we should be challenging the people at the top, and making sure AI stays accessible to EVERYONE. And not just big corporations. The goal of AI is to make things easier and more acceptable and I think we should be working hard to keep it moving in that direction.
We need to fight the corporations on this one. If it's ruled that "style" or general output can only be Owned by whoever owns the rights to the software or the rights to the training media, all they'll need to do is what they've already been doing: either buy out the AI company and put something in the EULA that ensures they have a stake in what's generated-- or otherwise have right to limit the output; or pay an artist to produce the art they can train the AI with as long as the artist signs away their rights to what they produce under the company banner-- and they'd simply outsource the creation to talented artists in countries where the cost of living is lower, as the corporations usually do, so they're not necessarily beholden to paying a living wage.
Once AI is accessible to everyone, corporations can't hold culture and entertainment hostage as they already do with immoral copyright laws. The people for whom it's important to see "people like them" on screen can make it happen instead of paying a "Creativity Landlord" for the privilege of being seen.
Thing is what are those people going to pay with when they are replaced too? And why put another strain on the wallet of those who need it the most instead of the billionare companies already siphoning all the wealth out of "the people"
Just so you know stable diffusion on automatic 1111 is 100% free. Open ai created it exactly for the reason to give access to everyone. Also the client base may shrink but your work also becomes easier if you are using ai tools.
Undercutting new artists from ever trying to get good. Early artists tend to need some hint at financial success and the existence of image generators will probably create a shortage of competent artists in a probably three generations.
@m f I don't see anything wrong with that. Markets change. And those brave souls left will become invaluable. Everyone else will still draw and paint for the joy of doing it not for profit. Jobs don't have to be your passion. It's sad yes but times change all the time.
I don't even think AI art is as great as some people seem to think it is. At least when it comes to the art I care most about, most of the time when you look too closely, you realize it's an eldritch abomination of some kind and during the few times when it's at least passable (usually when the places where it has the most trouble are hidden somehow), it's always really simple and in the same few styles with basic perspective and no action or idea. Besides that, AI art isn't really capable of most of what people who commission art really need.
AI can stay, there's no reason to argue against it because it's already here and isn't going anywhere. But I also can't see it being used for very much.
it will get better, the hands and the slightly uncanny eyes and the illegible text. Unless the tech hits a significant plateau for some reason in the very near future (unlikely considering renewed massive interest and funding), these aspects will all be improved upon quickly
@@andersonrobotics5608 Can we really be sure it'll happen quickly? AI can easily hit a significant plateau.
For example, people are excited for language AI, too (things like AI dungeon, dreamily, NovelAI, CharacterAI, etc) but it has been struggling with the same shortcomings for years now, or at least, the applications of it that are available for me to try have.
Some of them have made small improvements, while still conceptually being flawed in the same way, which is that each of them have no true memory and rely on a big wall of text for context, instead. This wall of text becomes larger and larger over time, which saturates the context and confuses the AI, before meeting a maximum character/token limit and being truncated, causing the AI to forget important details while also making up new ones that usually don't make actual sense.
When it comes to image AI, the problems we're seeing now have always been there, because AI doesn't know how to compartmentalize its data to specific parts of the human body and refer to them while drawing. The AI also can't tell when it's made a mistake and go back and fix it. But even then, the human body is only the tip of the iceberg when it comes to all of the things the AI has no idea how to draw, and especially in different artstyles.
It will try anyway if you let it, though, so we end up with people carrying living, soft guns with multiple magazines and weird gun-like shapes sticking out of it, or cars walking along the road with legs.
Which is pretty funny or disturbing depending on how you look at it, but I have no idea how it would fix these issues since there are an infinite number of different objects that can appear in a picture in just as many different perspectives or shapes or kinds. This problem, as well as hands (or feet or faces, etc), could easily become the plateau, similar to language AI.
@@andersonrobotics5608 So far, that is just an assumption, and we cannot make a judgement based on assumptions. Will AI get better? Definitely. But how much better? Even at the level it is today (which is underselling how impressive it is), it has very obvious flaws.
@@ShiaNeko I agree
@@ShiaNeko you say that the AI doesn't know if it made a mistake or mess but that doesn't mean it won't be used very much, simply that it will go through multiple iterations and humans will be judging and tell the AI where it made a mistake.
I think text generators or language models is fundamentally different from image generators so the growth of the technology won't be similar. This is the difference between Qualitative progress vs Quantitative Progress; quantitative progress is something you can graph on a plot like computing power but qualitative progress like image generators and language models is impossible to plot on a graph because we don't know the limits of the technology and if the problems are against the laws of physics or simply learning a technique to solve it.
For language generators we are improving but just because image generators and language models are run by AI doesn't mean the problems are the same. Just like batteries/energy and computer technology did not progress the same rate.
This is one of the rare videos I'd like to give more than 1 Like to!
“AI makes art generation too simple and allows even ‘unskilled’ people to make images”. I am sure the same was said about the first cameras and later digital cameras. Does that mean you would like to argue that digital photography is NOT an art form? If you agree that digital photography is an art form, does that mean that every selfie, insurance estimate pic, and accidental pocket snap would be considered art?
I think the moment you start arguing for automating art, you should realize that it's very anti-capitalistic to do so. The reason is simple: The automation will improve and there will be the point, you won't be able to make any money from your art. Because it will replace your craftsmanship the same way factories have replaced many workers in other jobs.
The reason I personally find going this route disturbing is that artist already have a very tough job to live from their work. Because art is subjective and art is not really something people can order. The AI messes with both of that things at once because now anyone can order whatever they want. Today an algorithm can paint an image, tomorrow it can make TH-cam videos or TV shows. Technology usually doesn't improve in the speed you prefer. So when you are going this route, expect it to go nuts in every possible way.
Because you might think, you are not just a creator of art but also something like a teacher on a video platform. Well, I don't see why AI would do art in the future but stop right before educational content. Then of course we can think of the benefits. Everyone could get access to that, right? Sounds good but I would expect that any algorithm will be built behind a paywall from a company. So pushing those algorithms will likely result in monopoles of art as well as other content and people doing anything comparable the manual way will suffer from that because they can't keep up with AI.
Honestly I personally don't think that it will bother me as much because I learned computer science. I know how to write such algorithms automating things for me or others. I expect to be able to keep up with this. But anyone who thinks that those algorithms will simply stay tools the current artist can use to improve their work is delusional. It would be one thing if the AI was trained on pieces from artists who actually sold the right to those companies using the AI. But from my research that's not the case.
Of course all of this wouldn't be an issue if we wouldn't live in a world where everyone needs to make money to survive. But that's how it is at the moment and I can only recommend artists to start looking for a different field of study. Because why should people come to you auto-generating images when they can do it as well? Sets you up in a very bad spot for negotiation, doesn't it? Bringing AI to the table is not a new tool, it's a competitor who can steal experience from others in any form or scale. Good luck competing with that.
When everything gets automated for the few rich who call the shots in this world, I wonder if there is any use for us common people.
Half way through this. I love your videos and I have watched every single one of them. But I think you need to watch more of Steven's videos, he is usually never this critical in fact... never, you can tell it comes from intense frustration of watching things unfold hurting a magical world he loves. I have many friends and students I was mentoring who were studying art and suddenly gave up learning to draw when they learned an AI can do it. A depressing amount.
I am somewhere in the middle I love AI and use it in my day job, but I do fear for young artists that feel like they need to use an AI to be competitive and may use it as a crutch. Like an aim bot in a fps, it would loose all fun. Every artist wants AI to take tedious tasks like rigging and weight painting. But the act of painting and designing stroke by stroke is a blast! Be like telling dancers that are lucky enough to be paid for dancing after working for years on it, that their dancing can no longer pay their bills unless they become a choreographer.
I think eventually people will be able to rekindle their joy of creating art, and be able to differentiate the experience of creating art compared to a robot creating art. An example would be there are still many people whp enjoy making jewelry, even though a machine can create the same ring or necklace. I think society will naturally value "hand crafted" art over AI generated art so maybe we don't have much to worry.
46:05 thank you for saying it. calling this stuff out is the hard truth that needs outing.
Props for the hard effort put into this well commented and personal presentation of the comments made by the creator of the other video. I didn't find a single point where you told people how they should 'feel' or tried to leverage the discussion. As someone who's spent years trying to make art of things I conceived and failing for a multitude of reasons, I find Stable Diffusion a complete life changing tool for my own gratification.
Let me be completely clear that in the past I have commission artists, thousands of dollars over the decades to render in some fashion my concepts, most did amazingly well, a few never completed the work and ghosted me, and many needed to redo their work because specified components of my request were not present, poorly rendered or just not in the spirit of the piece. Most complied but some told me that's as good as it gets for what I paid, "they have to make a living after all."
Then there's the ideas too far afield to ask someone to create. Stable Diffusion has managed these with surprising results that in some cases where better than the original concept or pleased me well enough I felt the task was completed and that was that.
Would I use an AI picture to help provide a reference to an artist to have them craft it to my liking? Absolutely, it would give them a clearer idea of what I sought and even provide details, words are want to describe or omitted.
At no point in our current AI technology, and quite likely far into the future; if ever, would an AI be able to write this kind of a comment, I feel safe to say, where the presenter in the other video feels sure now AI could render as good or better. There will always be aspects of art, in whatever form it takes where people will be able to do better than an AI.
Tools like Stable Diffusion are tools, people create tools to make a job easier, allow things to be done that an individual would not have time, skill or drive to do, etc. No one ever said, "Don't use a lever to move that stone, hire my brothers to move it for you, they need the money." Tools generally are created using pieces of ideas other people had.
Over the span of 5 weeks, I've used Stable Diffusion to make about 200k images, out of all those images I've had 7 I thought qualified as meeting my sense of what was artistic and suited my concept. Imagine if I'd paid a fleet of artists for that many images, how cheaply would they have to work for each image to fit my budget, how would anyone else find a artist to make their dream become a vision?
Already said this to Shad, but thank you for breaking the VERY one-sided narrative against AI art tools, Royal Skies.
This should be a discussion, but one side has turned it into a moral imperative dogma that can't be questioned, unless you're a "bad" person.
The dogmatic echo chambers/panic mobs online get on my fucking nerves.
Thanks for this video. It is nice to see people that can think logically and clearly. And accepting that others have a different opinion than you do, and that is fine, is rare.
Personally, I can really only think of two issues with AI art. I myself use stable(waifu :)) diffusion for inspiration in my own pieces and it works great. My first problem is that AI art will incentivize people to stop learning about art. Instead of taking the time and effort to learn about things like color, composition, lines, etc, people will just type "masterpiece" into a prompt and get a perfectly good image. I don't see the generation by itself as a bad thing, I just don't want the knowledge of art making to be phased out from society. As an anecdote, I want to make an animation with original music. As an animator, I already have the knowledge to do the first half of that equation. What I don't know is how to create music. I want to learn how to create music, but after learning about the development of dance diffusion, I no longer have an incentive to go out and actually learn. I don't want this to be the future, where nobody has any clue about the fundamentals of art(including music and everything else I might be forgetting) and instead just type a prompt into what is basically a search engine. In this video, you talk about how you want to spread your knowledge to everyone, but what happens when that knowledge is no longer relevant? I can't see any solutions to this, or if this will either be a problem, so people much smarter than me will probably have to discuss this. My second issue with AI art is the theft of art style. Just recently, after the death of legendary artist Kim Jung Gi, someone made an AI model that can perfectly copy his artstyle. This left a bad taste in my mouth, as well as many others, as an artist's style is basically their identity, and the AI has turned that into an algorithm. I have no problem with using AI and attempting to recreate the style of an artist you admire, as that's just what humans already do. I have a problem when you take hundreds of an artist's works and load them into a machine to pump out thousands of cheap imitations of that artist's work, especially if you claim them as your own. As to why I don't mind hand copying or even using an AI to copy their artstyle, you still need to input your own ideas and are using their style as inspiration. I don't think I need to explain how describing Kim Jung Gi's style perfectly to an AI and using aspects of his style to improve your generations is much different from training an AI to specifically copy that artist. If these two issues are addressed, I would be much more comfortable with the upcoming AI revolution.
@OliAnims
These are interesting points because the discovery of computers has opened a way to more sophisticated information processing that does lead to style, form, and expression, the kinds of things we identify ourselves as. But now, the question becomes much darker: If style can be accurately reproduced, does that imply it is merely data in our DNA that enables each person to present a style, or is style something more 'personal', whatever that means? Also, if style can be 'observed', does the Universe have a style which we can characterize? Or does style exist only in the information itself, and that is what the AI is extracting from the images (that is, patterns), and not seeing style in the image itself?
The second issue is whether the AI can replicate the person's personae well enough to create a history that isn't real, such as crimes and other nefarious actions, basically identity theft?
It may be the impetus for us to re-examine ourselves and our purpose as humans, life, and biology, rather than reject the technology.
Interestingly, we have turned full circle. The very copyright laws incentivizing people to innovate, learn, and progress are being bypassed and replaced with knowledge already gained through copyright. So, do we have sufficient knowledge for machines to do that for us, and we instead repurpose ourselves to more important activity? Almost every program written is imparting human knowledge into a machine so others do not have to know how to do the tasks. The AI is no different except that it is addressing the knowledge space itself, drawing patterns and inferences we, as yet, cannot.
@@noreply3322 "AI draws for you."
What do you mean by draw?
The AI is a program that samples images and reorders the samples into shapes that is applied to an output image. That isn't the traditional definition of 'draw'. The program links text words to visual shapes and those are linked to areas in images that the AI can sample and 'reproduce'. The AI composes images after 'learning' how millions of images have been composed. There is noting inspirational or passionate in that, nor creative. And by definition, Art is inspirational, passionate, and creative. The result is amazing and often stunning, but the AI isn't drawing for you, imho.
If your friend is the one that drew the cat, and the AI copied the drawing and learned how to 'draw' a cat from it, is it your friend's cat that the AI 'draws', or its own cat?
@@petergostelow I personally believe that an artist's style is a culmination of everything they've experienced in life. It's not just a mashup of all the art they admire, but also of their personality, their childhood, even something as simple as what they are for breakfast can have an impact on what they draw and how they draw it. AI, and those who use AI to copy, don't take any of that into consideration. They take the surface level patterns that an artist likes to use and make a mockery of everything the artist is. Now, I understand that the person operating the AI has THEIR own experiences in life, but none of that is shown in an AI generated piece, especially if it was SPECIFICALLY trained on one singular artist. Imagine having to not only helplessly watch, but also compete with a bastardization of all your life's work.
@@petergostelow DNA has nothing to do with "artistic style", DNA is a hereditaty data storage, mostly responsible for storing protein structure data. Artistic style is not hereditary trait, so it has nothing to do with DNA. Stop using scientific language you barely understand. Artistic style is a combination of aesthetic preference and art training. Artist trains to use instruments to copy and modify existing art to their preference, after they master the instruments and art techniques it's just a matter of preference.
@@xn4pl "DNA is a hereditaty data storage"
DNA stores a replicating program. It is responsible for building everything, including your brain and how it is wired. Therefore, personality and characteristics are also embedded in the code. If not from DNA where does the personality and preferences come from when you are one cell in the womb? Do you know?
Each person receives about 1/2 their DNA from their mother, and the other 1/2 from their father. Each child has different parts from each parent so they appear different, but the same. The parents characteristics are passed on through the DNA, modified and adjusted, so the 'family' share common personalities, art style, and preferences, but none exactly the same, not even for identical twins.
Where does personality and art style come from, if not the brain, and where does the brain come from, if not the DNA?
If art style were not a pattern and procedure, it could not be replicated in a computer program, could it? Well it is being replicated as numbers in AI. Both the AI and the DNA are numbers arranged in a specific sequence. The AI shows human style from human output, but stores that style as a sequence of numbers, a different pattern for each person. Where did that pattern come from, if not the DNA that made us?
This video is criminally under viewed. I appreciate your input and criticisms. I also appreciate you calling out OPs pretentious bullshit. It’s true that I have a ton of ideas that mean the world too me and not enough time to dedicate to practicing art to make it real. AI has been an incredible asset for me in putting my ideas into reality. Where it goes from here, I don’t know. But I’ll be excited for the future.
Royal, I respect you and I've learned some good stuff from you. I feel like you try to come off as the voice of reason here but this is two sides that are both biased as hell. I think the other guy presents things poorly and sounds like an asshole but has some very good points. You yourself recognize the value of time and effort going into art and creations. A lot of your videos descriptions mention your game and how it took thousands of hours of heart and soul or something. Imagine this future where you instead say "Check out my game it took over 20 minutes of prompt engineering!" Nobody will bother, they will just make their own. As a channel that has primarily thrived on art tutorials, the easier it gets to create, the less demand there will be for your content. I do think there is no going back on AI and whatever he is trying to do to stop it is pointless. But I also really believe it will devalue art exactly as he says. I've generated thousands of images on midjourney etc and it was fun for a little while, then they start to lose any meaning because they didn't actually take any real effort. I'll always cherish my hard -earned blender creations but a generated image anyone could make in 10 seconds just doesn't hold that value. The future of art is looking really bleak in my eyes, everyone trying to be a "prompt engineer" sounds boring and hollow. Personally, I won't give a crap about peoples AI generated art and I wouldn't expect anyone else to care about mine. You think this world where our grand children spin up a michael bay film or whatever is awesome. And I think this world where our grandchildren have no hope of turning their artistic passions into a career is depressing.
Honestly, I feel like most artists, including myself are accepting AI for a Darwinian reason; either adapt to the new tools and new world and thrive in the competitive field of art to earn money or resist and die. Not because we're this excited about it. Personally, I empathize a lot with 2D artists. AI basically stole their talent and rendered them useless seemingly overnight.
That is some scary shit. Even writers, voice actors, and composers are now facing that threat. Shit, there are some AI tech now writing essays for students in college. When human effort gets this minimized, it's not a good sign. Who knows what the future holds? Personally, I keep an optimist point of view, but it's very hard to, at least, not be concerned with the threat of AI.
Personally, I find the idea of just writing what kind of film you want, pressing enter and boom you have it generated is boring, bland and wack as fuck. I can understand its appeal for NSFW content, but for everything else, human intent behind the work always makes it better.
We are all afraid for our own future, its natural because its unknown, unpredictable therefore terrifying.
There's no threat. If it's any consolation, people were talking about GitHub Copilot & Co. replacing programmers but that seems a bit laughable after using it for a while now (though it can help your workflow which is nice).
Though maybe programmers dealing well with automation isn't the best example.
Yeah. Even with the same tools, the average Joe with no artistic formation cannot replace a trained artist. AI renders the brush strokes and the muscle memory redundant, but little more.
@@SimplyVanis not really unpredictable, we know exactly how this ai will affect the future of art desu. It's not fear of the unknown, it's fearing about how much tech fucked up the future of art
You're setting yourself up for failure. There is no reason prompting won't be automated just as well. Artists that base their work on their personality and caer are more likely to succeed in this environment than prompters.
tl;dr: "we're at the sweet spot on a very large wave of excitement that will pass. We shouldn't make any new laws until the new-shiny of AI art has subsided"
I'll say what I've said elsewhere (haven't watched more than 2 minutes of the video so far, so I'm hoping you cover this): Right now AI art is in a "viral" phase of popularity. There is a wave of excitement that is going to have to pass before we should make any lasting changes to law. Millions of people suddenly have an outlet that makes them *feel* like the artist they've always wanted to be. It can be like someone just discovering synthesizers, they can make all sorts of new and exciting noises, and they are thrilled with results. They see a future of themselves as a popular musician etc. The problem is that this is still the initial "new shiny" rush of excitement. Often times people will drop money (lots of it) on new shiny, only to have it sitting and collecting dust a year later. Because in order to make the new shiny into a full dream takes a lot of effort. Taking AI at it's *current* state, sit down and try to get an AI system to make *your* vision for the image come true, and it's going to take time to get all the details right. Inpainting here, photoshop adjustment there, etc. Again, once people start to hit that "effort" wall, the wave will pass and *then* we should be taking a solid look at the legal ramifications.
People who are calling for massive legislation etc. *right now* are more than likely just opportunists looking for clicks and ad revenue, and it would be colossally stupid (or nefarious) of any government to make snap decisions based on what is happening right now.
The viral phase of the excitement and popularity needs to pass before making long term decisions. I'm positive that Adobe and other large companies are already looking into ways to monetize things as it is, and with that will come custom models that are "copyright clean", photographers and artists will be offered contracts which will grant them money in exchange for exclusive rights to using their images in models which, coupled with advancement in AI generation tools etc., will all be sold to users for a likely too-steep monthly fee, while AI detection will be incorporated into methods of stifling models that will be deemed "illegal" due to copyright claims.
Glad I commented before watching. Still have to say, I find it interesting how he constantly uses the word "humanistic". He's literally implying that people who use AI are not, in his eyes, human. He needs a solid psych evaluation.
Given how fair use is supposed to prevent people's work being used to put them out of their jobs, I couldn't disagree more. This was illegal from the start, the most important clause of fair use is how it affects the value and market of the original. The idea that we're just supposed to sit back as they pull the rug from under us doesn't make sense to me. I don't see the progress.
Great video! Thank you for taking the time to do it.
But, that is like you pay some one to keep your self safe from other people, but they guy who protects your stuff lets people use the stuff you have. kinda unfair
Here's the thing that I have come to realize. We can put in new laws in our countries aimed at preventing AI developers from scraping art and similar such things as some sort of copyright protection. However, those laws would only work in the sphere of North America and Europe. There are so many countries that don't even remotely care about Western copyright laws, and a cease and desist from us isn't worth the paper that it's printed on. Do they think that a Chinese company developing AI is going to care about American and European copyright laws? How about one from Russia? Or India? Or several countries in Africa and Southeast Asia? Pandora's Box has already been opened, and at this point, would only kneecap our ability to compete in technology with several countries, some of which are our geopolitical adversaries
The most hilarious thing that about that point Steven makes on art in an AI-dominated world not mattering is that one of his other most highly-viewed videos is LITERALLY titled: “You Should Not Care What Other People Think About You Art.”
Lmao. His whole video seems like it was made disingenuously and with the sole intent of farming rage-clicks from artists who feel threatened at the moment by AI.
Some folks use art as a recreational activity-they go to the lake, beach, whatever and sketch/paint etc. just to pass time or relax. That’ll definitely still be a thing.
Absolutely. But the artist that made a living off their artwork will be devastated by the AI art revolution.
The argument in the end is really toxic, honestly. It's totally fine that you personally agree with companies using your published artwork to improve their algorithms. But you can't expect everyone to think that way. Also I really don't understand why it should be okay for those companies to use your artwork for training, only because you use a cloud service to host them online which has nothing to do with any of that.
It's one thing to agree to the terms of service with Google for hosting your files and a completely different thing to agree on third-party companies querying your data to train neural networks replicating your experience and knowledge to sell. If they ask you for permission and you agree to that, everything's fine. But that is not the case.
Only because you think that people loose all of their copyright or ownership the moment they make a Google account, doesn't mean that's legally or ethically the case. It also doesn't matter for the whole discussion whether it's technically legal at the moment or not because it's not like there has been a final decision in legislative about copyright violations in content of training for AI generated content creation.
So hell no, artist should not be thankful that stable diffusion is providing an opt-out functionality. Because it should be a requirement to have opt-in here. That's the argument. That's the reason he points towards music generation where lawyers tend to be much more involved. The reason why artists are thrown over here and musicians are not is the difference in amount of lawsuits you get by doing this whole training process. It's not about the resolution of the content.
Hey, not disagreeing or trying to provoke in the slightest. But can you share your source? I wanna look into it and understand your argument better.
@@asshat8892 What do you mean by my source? I state mostly negatives here like "they haven't asked for permission", "there's no final decision" and others.
Have you ever tried to proof a negative? - I thought so.
The question is why should training a neural network be treated differently than a copyright violation in case you haven't asked for permission. In the video he brings up that he personally is fine to use services as Google drive to store his personal work. Then he continues that this implies it is valid to use/modify/copy existing artwork from others published via other services on the internet while completely different legal terms or licensing might apply.
It's not even difficult to find neural networks being trained on stock images which are typically sold online. So in other words you usually have to pay for usage.
If you then train an algorithm to generate pretty much comparable images without watermarks while avoiding its license, how is that legal? It certainly does not fit needs of research. It's just done to bypass copyright. Pretty obvious.
Personally I don't see an issue with neural networks being out there to be trained on images with permission by the creators of its source material. However that's not what we see in practice. A little bit of research on TH-cam and you can find multiple videos using copyrighted images, photos, drawings as source or target materials as well. The majority using this tech gives a crap about artists.
But then more people might benefit from it overall, right? You could bring the same argument for movie, game or music piracy. As long as there isn't a systematic financial guard for artists, I don't think AI should be allowed to circumvent copyright. Because otherwise who will make the material to train next gen AI?
If AI generation is legal, we would actually need to rethink copyright completely. Why would I not be allowed to make a 99% copy of X if my computer is allowed to do it faster?
Anyway my main argument was originally that he made a statement in which he implied his own point of view would apply to everyone else... and I think that is quite toxic however you believe AI should be treated.
My personal recommendation is downloading an open-source implementation, train it with your own artwork and generate stuff for your own use. In that case nobody should have any issue however they stand on this topic.
I already watched Steven's video, and objected to certain parts, but watching this now brings back the parts which really annoy me. In particular, 44:01. He seems to say that your ideas won't get attention, and therefore won't matter. As though likes on twitter is the only form of validation worth pursuing. His idea of the 'mega-feed' is in some sense likely, since all that's required is the sheer capacity to make the thing and someone willing to do so, but it won't be so all-consuming as he suggests. 'Not even your mother will be able to find it' unless I, you know, show her. Or send it to her. If all of four people in the group chat are capable of producing all the entertainment any of them ever need, and sharing that experience with the other three, I view this as a pretty good outcome. It just means that we must narrow our scope with greater capacity, rather than expand it.
The presence of the 'mega-feed' is not sufficient to force you to pay attention to it over a few individuals you actually care about; the only way that could be done is by wresting control of your devices and channels of communication out of your hands entirely. The creation of AI does not make this any more possible than it is currently. That is to say, almost possible in closed ecosystems like Apple's, if not for alternatives like FOSS. There already exist decentralized ecosystems capable of keeping this power in your hands.
I think most of the flaws in his argument stem from his apparent malignant narcissism. Love the piece he's working on...
Obviously I have a very limited exposure to him and my assessment is not a complete one, but his language reminds me of every narcissist I've ever known to be a narcissist.
Your counters are very well thought out and articulated respectfully, despite your disagreements on the subject matter. Great video!
Artists can be a proud bunch. On artist TH-cam channels that talk about AI, I see a lot of comments that suggest the poster is specifically upset about people using AI image generators potentially getting undeservedly recognized as "artists", while they perceive their own artistic achievements are devalued.
@@The8bitbeard I can understand that. I get annoyed when I see obvious AI art promoted as if it's a hand-done photo-real pencil drawing.
This is the most important video of the internet today. Thanks for your time and patience.
Steven just sounds insufferably pretentious, arrogant, and emotionally manipulative. I can't listen to arguments made by people who are clearly projecting their fears (worse presenting them as valid arguments) and who seem to want more to be an author than a visual artist.
As a self-taught life-long artist who is messing with AI, I can say I haven't been this excited about creating art since I was a kid. It is massively inspiring to me. I also have no idea how someone is going to get really unique generations without artistic knowledge. It isn't as simple to generate an image you have in your head as you might think. It's better to go into it without a vision and adapting what it gives you.
Adam duff actually made a healthy argument in a video about where the line must be drawn when it comes to AI, and i think he makes such a great and valid points, i recommend you check it out.
Yeah, Adam's take is MUCH more defensible than the gatekeeping ridiculousness of SZ's video.
@@done.6191 Bullshit. It is indefensible to use a person's work to put them out of their livelihood. THAT is indefensible, not what Zapata said.
@@Real-HumanBeing are you in the profession? Because no one is being put out of work here.
@@done.6191 You know that how? I've heard of many artists that said they aren't getting comissions at nearly the level they used to, and the tech is "in its babysteps"
@@Real-HumanBeing I should be more precise I suppose. I'm admittedly talking about a specific subset of artists, that being artists working to fulfill specific briefs from companies. I'm NOT talking about people drawing furry porn for people or the same Anime Waifu a hundred times at $10 a pop. There is nothing at ALL wrong with doing that, but It was never a long-term job position for more than a very few people.
Source: professional artist of 25 years, making money from companies that pay me to do it. None of them have even whispered about AI art being used in any way other than as a tool, which is how it has ALREADY been used for the past 3 years+.
You’re too kind to them. The amount of condescension and entitlement out of Zapata is at critical levels. We’ve gone through this song and dance so many times with evolution in technology, even in the arts. I’ve fallen for it before myself. I used the “canned music” excuse when garageband became popular, and I was wrong. Almost none of these arguments against it are new.
As someone who turned to 3D because my hands are too shaky and I’m too busy being a wage slave for corporate to really draw or paint anymore. Pulling AI into my workflow has been so invigorating.
If this really does take off I predict it will either oversaturate the market with CG images to the point where no artist can make money from selling the art and thus balancing the system as the hype dies out and the softwares devolve into artist tools
Interesting statement. I'm a little older, but I've seen the traditional art market become less and less relevant. I think that AI art may quickly make traditional art become relevant again. Things like Gouache, watercolor, acrylics, and oils might be skills that artists want to study.
The argument can be compared to a hiker and a passenger.
The hiker hikes for the journey, not the destination.
The passenger is only in it for the destination.
To the hiker, the destination is the final reward. The hike is a story unique to them.
To the passenger, the destination is not a reward. Just something to cross off the list.
The hiker will experience the cold winds, blistering heat, and so much, but will be one with the world.
The passenger is protected by the carapace of the vehicle.
The hiker can arrive at 10 destinations and will feel fulfilled.
The passenger can arrive at a hundred destinations and will never feel as fulfilled.
The journey has a destination, but is it really worth it to cut the journey out to instantly get gratification that might've been more fulfilling if you hiked the road yourself?
You are a passenger, you are in it for the destination. And the vehicle is too fast to fully enjoy the vistas.
And if AI is to be a vehicle, it is the fastest vehicle. So fast, that the only window in the front shows the destination. The journey is negligible, the vehicle is too fast for the journey to even matter.
I love your 3d tutorials for their destination oriented methods.
But if it is about art, it's better left with the journey so that the destination is fulfilling. Calling the journey a waste of time, is a waste in of itself.
Your philosophy on getting to the destination fast feels like it only applies to coding and technical stuff,. Like they are actually guides on quickly learning how to "walk" in order to go on a hike. But the direction you are going... Nope, I will not follow you to your super fast vehicle because I want to enjoy walking and you have way too much strawmen. The draftsman show recently started again with this topic actually getting professional voices on it.
But say what you want to say, I think AI collages makes human art more important than ever.
Like, "huh, this is made by AI?" and "wow, you didn't use AI on this? Amazing!"
Edit: I still love your 3d and coding videos. I think you should just stick with them.
Yeah some of his arguments just infuriate me. I am not an artist and a philosopher, but I am an artist and a philosopher. I am not the former because those are not my professions and I don't hold degrees in those fields and I don't make a living off of them, but I am the latter because I have created new work from my imagination in various ways and have pondered and discussed the mysteries of life and existence. Just because I am not publicly recognized and money is not exchanged doesn't mean that my work and thoughts can't have value, to me, my friends or some random people on the internet.
I know some people that feel they can't write the stories they want to write not because they think nobody will read them and won't make money, but because they can't live up to the standards of some random people on the internet and because they fear those who complain about everything and generally as groups are vocal and have opposing views on various subjects, so they are stuck because they care too much. That frustrates me a lot.
Why should we care about what others think or want in order to create what we want to create or how we create it. The elitists can just shove it. Why should they belong to some privileged class that can't be disrupted by new technologies and other events, everyone else is exposed to this risk.
I clicked here atracted by the "thumbnail" sayng "healthy debate"...
its not a debate when the other side cant reply directly to you.
This is the reply.
@@froztbytes yea, this is a respond video, not a debate
There's one thing I think you may have misunderstood in the section at 45:00, your biggest disagreement. Steven said not that the dream anime is worthless, but that in a scenario where art is effortless, and therefore the supply is limitless, that no one's art will be able to get a foothold/a strong share of attention.
You won't believe it but even within human only competition most of worthy art doesn't get attention it deserves, and in cases where it does get attention it's mostly because artist already has established their brand over the years. Also limitless supply of art is really a pro and not a con of this situation, unless you're 1 percent of population that required that to not be a case to extort people of their money.
@@xn4pl I agree with your first point, actually.
@@xn4pl I'm not sure doing a job to try to feed yourself counts as extortion. People's jobs are being upended, but their job isn't an extortion racket, anymore than a plumber fixing the pipes in your house and wanting to be paid a living wage for their effort is extortion.
Watched yours, Jazza's and Shad's videos regarding this topic.
My concern with AI art has never necessarily been that it "steals" art from artist. But rather two different things.
To keep it brief; rather that as this technology improves, and its used by major companies or future "artists" the unique aspects of artistic mediums will be boiled down to an algorithm. Things will become "samey" because that's what people like as a whole. As someone whose hobbies tend to be artistically driven (yes much of it using computer assisted tools, yet still deliberately chosen by the creator) I don't like the idea of future media being driven by a formula and the potential homogenization that will bring with it.
Not only that, but its hard for me to see AI art as something that was "Created by an artist" in the same way. It would be like me paying an artist for a commission, asking for different tweaks and changes as the piece progresses, and then claiming the final product as "my art". Yes I may have the ability to perfectly envision what I'd like the image to look like in my mind, but not the skills to manifest it. Telling someone who does have the skills to create it all the minor changes I want them to make until its made manifest does not make me an artist, but more akin to the "idea guy."
The artist made the work, I told them what to make. AI created art, in my opinion, follows a similar path.
I'm sure theres small uses of AI to improve ones art that I'm okay with, such as maybe tweaking a pose, or coloring here and there, but there also comes a point where I don't really see it as "you" doing the art anymore. I'm not sure exactly where that "line" is just yet, but working on figuring it out.
I would not consider myself an artist however so maybe my opinion is off base, but thats how I see it.
I come from Shads video and i actually agree with you on the latter paragraph. I consider myself a digital illustrator, and im currently studying graphic design. I currently use Photoshop to do my art pieces because of the number of tools it provides: transforming an image, levels of brightness, the crtl+u that opens the Hue/saturation/brightness tool and allows me to twerk stuff quickly... Its amazing. And i do see AI´s enhancing my work. EG. I had issues this year because a stock picture that i wanted to use for a work of mine wasn't long enough, and i needed to "invent" the rest of the image by using tons of collage, tampon and my graphic tablet to fill and make the image seem more coherent. If an ai can do that tedious job for me, im all in.
But when it comes to art, i still feel Ai´s although legally allowed (because we all had to suck the terms and services of any social media) aren't morally correct yet. I´ve seen Shads process when he did his wife, and the Armour design isn't his making, and if it were me i would not claim i fully did that piece of visuals. It feels like a cheating way bigger than the ones i could do on photoshop. I wanna embrace ai, but i fell so nasty for using a bot that serves me everything done
As a person who suffers from aphantasia, I have always struggled with lighting in painting among other things. I never post an AI generated art piece on my main page and call it my own, I always work on it quite a bit. Once I have the lighting figured out, I usually draw over it, making it my own. I am not against AI but it does need some safety protocols like maybe you should not be able to generate realistic images of regular people and put them in a crime scene for example, this scares me. Using it propaganda pieces, I dunno... not everyone out there has the best ideas in mind. Look what happened to Facebook and Twitter, people now use it only to spread hate and propaganda. I fear AI is capable of making things worse. I do enjoy the process of ideation with AI, its given me ideas that I would have NEVER thought of, otherwise.
With a condition such as yours, I totally empathize with the usage of such AI generations. In fact, I think it's so important to have these conversations about artist's rights around AI specifically because of people in your position. The current AI models have a large number of unanswered questions about legality, ownership, and what level of input by a human could place it under the "Fair Use Under Modification" clause. For me, that's one of the main reasons I'm so against this narrative of AI Vs. Artists that is being perpetuated by both sides, despite both groups claiming that "WE'RE THE REASONABLE ONES, IT'S THE OTHERS THAT ARE MAKING THIS AN ISSUE". With licenses such as CC0, CC 3.0, and of course the plentiful number of items in the public domain, there's a clear-cut way to make an algorithm for people in your exact situation without having to even ever touch a website like ArtStation. Typing in "Mario Kart by Simon Stalenhag" or "Seinfeld in the style of Banksy" is fun, but when you start commercially selling those products with the intention of mimicking that artist, then suddenly there's a big free-for-all at the fourth clause of Fair Use, and everyone suffers for it. It's not about removing the AI, it's about making sure the AI protects both the artist whose content it's trained on and the artist who might use the algorithm for inspiration.
And that's how I think AI generated art should be used.
Sadly many people don't understand that or utterly missrepresent/missread copyright laws to not think about how AI assemble art
@@masterzoroark6664 It's not about how it assembles the art, it's about the fact that a machine (one that's going to do a way better job of learning patterns and reproducing things because that's all it knows how to do) is being trained on people's images to learn. It still has to use the image at some point in the process to learn, therefore at some point it becomes about unauthorized usage of images for commercial purposes, thus the argument still needs to be had.
@@mbishop3d832
yes, I know and stated that, sorry for missunderstanding.
We just probably need to wait till one of these generators spits out pracitcally Disney or Games Workshop merch for people to really notice they can't hide behing "Machine did it".
Tho on the artstyle point- so far AI showcased it's unable to make a style of their own. If you ever drew or saw someone drew you know that with each attempt to draw like your idol you still will draw differently. That's how artstyle is formed, and only disingeniuous forgers don't ebrace this fact.
I am of belief that once AI makes artstyle of it's own and artwork from it's own creativity then the artwork is it's, not the person's who overseen it create.
@@masterzoroark6664 Good points, also no need to apologize, I was just engaging in the conversation with you!
As someone who regularly needs art commissioned for my musical projects, I've hired through places like Fiverr and artists I've known personally. With the exception of one artist who I could literally walk down the hallway and look over the shoulder of, almost all of the art I've ever commissioned was discovered to be photobashed or outright stolen, despite paying hundreds for it. I was able to backtrack through and learn that EVERYTHING was always photobashed or straight ripped from other copyrighted sources, even though I had paid for commercial usage rights and spoken directly with the artist. In my attempts to support the independent artist communities and paying multiple hundreds of dollars for individual pieces, it's only through AI art do I actually feel SAFER from copyright infringement and like I'm actually realizing my vision. There's also an argument to be made that for musicians like myself who operate on very rapid release schedules, being able to sit down in a single day and get the art I want and pair it with my release is a tremendous value for audience, as I can deliver faster and with a more compelling image to accompany my music. I can say with confidence that AI art has never replaced a commission for me though. I would have made the art myself (like I had for years), but now the quality level of what I can do is far and away superior.
Just check this guy making drawings: th-cam.com/video/5ahGvcC2_HQ/w-d-xo.html You were paying to the wrong persons
And Your story also illustrates how hard is for real artists to get jobs and now thanks to AI this situations is worst
You make a good argument for AI art. I'm sad but not surprised to hear about the rampant photobashing and stolen content artists are selling. Fiverr may be part of the problem. Artstation seems like a place for more reputable artists but wouldn't work for inexpensive, fast release art. Frustrating but a good case to learn AI for art.
@@The3Dsmash Fiverr definitely contributes to the problem, since it rewards fast, cheap, and easily produced art. It does feel like it's a "pick two: fast, cheap, or quality" type of deal. But on the other hand, I've hired artists for illustration of comic books and had a far more pleasant experience, so it seems to be very much a crapshoot, haha.
@@DaveyPerron "pick two: fast, cheap, or quality"
Exactly! Good luck with your music.
To be honest what he finds as a horror at around 49:00 I find as a dream. Where it finishes my lost or trailed thoughts. Is a helping hand getting past my creative depressions. Ends up being a tool to help change my perspective on life while still being tailored to me. It would be so cool to have ai movies generated just for me to consume so I don't have to wait months for the next A24 movie to drop which already seem like movies tailored for me. I think it is crazy to believe we wouldn't want to see a world that revolves around ourselves as most of us already seem to do. Human interaction should be a choice and not mandatory. Replace workers with robots and machines. Art with generated components be it filtered through meat or metal. Creating the next step of humanity and making something more than human is what is (in my opinion) the natural progression of the human condition.
@Royal Skies. I love your mentality and that you (like me) believe we should make things easier for others. That just because you spent so much time on something doesn't mean you shouldn't give someone else a short cut. Plant your trees even though you might not ever live to see their shade.
@@TheCaptain610 first of all I run off the theory (which I have yet to find anyone else talk and is something I'm building) what if God built us to be better than him obsoleting himself. Granted being an atheist I counter my own theory but it is still in the works. Though nobody would be billionaires because like in our world no money is completely baseless though would actually have zero need. People who aren't in that upper class range if money does still play a roll in the economy would do whatever. Yes robots do everything but it doesn't keep people from doing anything. People can still pursue arts, maths, sciences, or stem jobs like they do now. Just there isn't the need. It is a want. That's the biggest difference. And something I think literally everyone overlooks in an automated world.
@@TheCaptain610 Whatever a person wants to set themselves as. If UBI becomes a thing, I see it simply as the ability for people to find what they truly want. This can either be someone whose nothing more than a consoomer, or someone who decides to explore and become knowledgable in very extreme niches. The ability of people to be able to live for their own desires without a job taking up the vast majority of their lives just to put food on the table is a net benefit for humanity.
@@TheCaptain610 I didn't really elaborate on what my ideas for how to make UBI would work, which I apologize for that. My simple idea is that the gov would have nationalized space/moon mining that they would use the resources from to fund the UBI. I do not trust the CEOs at all to do it. Of course, this would require the government to have some actual changes to address the corruption currently in place to actually facilitate this (boomers are still in power after all), but I think this is the most realistic way to achieve UBI, and I don't think the government will just get away with trying to let mass amounts of people starve as automation becomes more widespread. It's one thing to have the majority of the country living paycheck to paycheck, another for them to be unable to eat.
Man, I follow you for about 1.5 years probably, started from interest to BNPR-render-blender stuff, and really like your attitude and style and determination in previous videos.
But this one come out too protective, which means yourself was kinda emotional. There were some slightly better arguments towards the end of the video, but I'll say about appealing to Google and other's big tech Terms of service is transcendental level bullshit. It is kind of way of saying: "Well you uploaded your pictures to internet and taken photos with your phone, they'll be used and abused anyway, feck you bro." This is bringing up immoral legal loophole, which should not exist by common sense, but is present to serve corporations, because they are and always will control legal field to their favor.
Also I am a professional 2d artist and at recent times used Stable diffusion super heavily, and got awesome and diverse results from it. And this did not made me happy at all, for variety of reasons. (allright, nobody cares about my tiny puny artists feels, no problem with that)
But we as humans are loosing focus from real quality, struggle for personal growth and improvement, to quantity and superficial approach. We get more and more possibilities to do stuff, but games and movies for weird reason do not become better, quality deteriorates. probably around 80-90% of top notch high quality art from artstation, for example, is already looking the same and boring. Everyone is repeating everyone else, and now AI comes to play, and for now it is awesome new toy, yeah, I enjoyed playing with it too. But over time people WILL lose their personal voice, and only one who will benefit is provider of AI.
Also call me the tinfoil hat guy, but I do think that going opensource was a smart-ass marketing move, no more and no less.
And also you know what, I would love to be wrong on this, and I would love that Steven would be wrong on this, and everything will just become better with it. But knowing life and dem peoples, most probably it will not. We'll see.
agreed
"No attention, no clout, and no money"
Super telling here. Some of us make art for its own sake, not for profit or prestige.
I think that AI will replace artists, because why should you pay an artist if you can get the desired picture for free.
2D artists are already almost completely useless. (only animators are still needed, but not for long).
And the bullshit that these AIs will serve as tools for the artists, and that the artists will be needed to work properly with the AI, is nonsense.
Artists simply won't be needed, because it's incredibly easy to learn how to work with AI, and the scenario in which a person will be paid to enter a description of the picture and click the "generate" button is just as stupid as possible.
The truth is, that artists will very soon lose their jobs and that's it. No alternative jobs related to working with AI. No happy ending.
There is no longer any point in artists job. Everything is simple.
AI cannot replace artists, if artists become extinct in one day, then AI too. Because AI will not be able to train without artist work on the internet.
Do you know Vocaloid? The company that created Vocaloid hires voice actors to and stored their voice in a voicebank and then makes a synthesizer that can sing and make music based on the voicebank. Did the singers lose their jobs? No, they didn't. In fact, they got more work because of the trend of covering Vocaloid songs, which is intended for people who like vocaloid songs but don't like the mechanical robot voice.
And now many music synthesizers use AI Voicebank, in order to get realistic tune like humans. Did many people lose their jobs? No, they won't. Instead, many people who understand music but can't sing will become music producers using the synthesizer.
And AI is a tool for artists to improve, especially since AI images are copyright-free. Artists can improve faster with the method of training by copying without having to fear being sued and made into drama on twitter
@@linaqruf AI images are not copyright free. You will see people calling it AI-assisted art
I don't think it matters whether current AI training sets contain little to none drawn images or whether it is mostly based on photographs. It's still violating copyright as far as I can tell and the set of weights for the neural network couldn't be calculated without the data.
I fundamentally disagree with your vision for the future described at 6:20.
“Harry Potter lightsaber red ranger adventure” or whatever already exists, it’s called a Content Farm. “Among Us vs FNAF vs Friday night funkin at Squid Games Spider-Man Elsa etc etc.” and plenty of iPad babies are already frying their brains on it.
You call it awesome, but I don’t think the whole of society’s entertainment should be the cinematic equivalent of jingling colorful plastic keys.
Even if AI got better than that and was able to make content that seemed “meaningful” on the surface, that is still largely by accident. I would rather experience art that has meaning on purpose. Your bullet hole analogy ironically shoots your argument in the foot in this regard.
Then there is the ethical dilemma of how many people would be put out of a job in Hollywood if studios starting implementing these technologies and replacing employees. Suddenly thousands upon thousands of people are out of a job in an industry killed by AI so they can’t apply their skills for gainful employment anymore.
In the ideal future presented by this video, a slew of struggling artists and unemployed writers and animators sit in their decrepit apartments as society looks on at Content Farm content presented as mainstream, devoid of the humanity, soul, talent, and hard work that makes art worthwhile at all.
That is Hell.
You should watch the movie "Tim's Vermeer", A technician analyzes Vermeer's work, and figures out how to draw a Vermeer painting without any painting skills. It shows that even old masters wouldn't hesitate to use a new technology to create their art.
Regardless of what happens, creative AI just sounds like a dystopia for me.
The only argument you put forward that I really have a strong disagreement with is the concept of consent. Where terms and conditions that are written full of legal jargon that would take even the best lawyers time and focus to pick apart, the fact that most people hit 'I agree' does not, and should not, automatically mean they consent. A company cannot for example just insert a clause stating that by using this software you agree that the company can at any point in the future demand 100% of all the revenue you made from using this software and expect it to be upheld by the law.
Please make more long videos like this, I love your effort and logic put into these. These are a great listen.
seconded
+1
When everyone's an artist; no one will be
theres an artist in eveyone my guy.
@@peacefusion Not if you don't really care about it.
I use AI for writing because I have limited use of my hands. They are my ideas and I am directing the flow of everything. I still do a ton of editing, organizing, fixing and rewriting though. It's still a ton of work. When I had use my hands I would have never used AI and been on the side of the artists going against AI. However, after I lost use of my hands I thought my life as a author was over, it took too long to write anything. AI has helped me get back into the saddle and my life has meaning again.
46:30 this rant. so much yes. 😂
Why would I ever buy ai art from you let alone commission you when I can do it myself?
It’s important to note that even though the current data collection practices are technically legal, copyright for AI generated art isn’t. There is already precedent for this, so as long as the law don’t change, artists like myself just cannot use it commercially and professionally at all, all you can really do right now is generate and share on social media, which will make your art and it’s idea open to plagiarism without much legal action available to take. Not to mention stable diffusion right now still isn’t robust enough to generate consistent images, it really struggles with recreating the same designs at multiple different angles, without changing the fundamental design all together, and it also doesn’t allow you to generate images in layers yet, so you also can’t really change the lighting or local color without again changing the whole design all together. We need better AI design that give you more control than what we have right now, that’s the only way for it to truly become a tool for artists instead of a cheap way for normal people to generate random images, because artists don’t care how good looking the final result is, they only care if it can be easily controlled by them and create the way they intended, instead of an accident, and robust photoshop layers, clipping masks and blend modes with strict mathematics behind them will always be more robust than neural networks based on probabilities generating images on a single layer.
One of the best videos of the year, SuperEyepathWolf's "Influencer Courses Are Garbage", he talks about the joy of creating. About expression for it's own sake, and delivers the powerful message:
"There is nothing wrong with you if other people don't pay attention to your art. That art is for you. That is enough. I promise."
According to Steven though, your art is worthless unless people notice it.
I am not going to attack him for his views, but that, his elitist, gatekeeping attitude, and his constant appeal to emotion are REALLY rubbing me the wrong way.