In my opinion your solution for this ai ethics problem is forgetting one thing: these learning machines are made for profit, they are not made for the artist, the are made for mass consumption (that's why their prices are dirt cheap). Just because one person in the ai field said that "ai is not replacing artists and that it is a mere tool", does not make it true. Huge corporations are the ones investing billions of dollars, they don't need artists since they got their artworks for free. So this is a very profitable, very eye-catching business model. You invest in the tech, you invest in a non-profit that gets you 5 billion images as a database then you use prompters as a feedback tool for the machine (free work and sometimes they pay to work basically). It's true that we as humans are to blame for unethical or bad use of the tech, but that does not excuse the tech bros that promote "ai is democratizing art" speech while destroying the market and art communities. And I swear the "ai is here to stay" line is getting reaaaaaaly old. It only discourages people for fighting for fair use and steers towards complacency. Ai as it is now is not ok, and yes, it's the opinion of the few atm. But tides change.
It's true that in this capitalistic society, AI art is very profitable and that's why corporations will jump on it. That's why people like us are here to raise issues about it. Also, I disagree that "AI is here to stay" discourages people fighting for fair use. I do agree that some people have adopted a nihilistic approach, ex. "AI is here to stay so what's the point." However, "the point" isn't about the end result, it's about the process of how we get there. And it's up to all of us to educate other people to understand it's not about the end result. It's about how we get there and how we set up a good foundation to support everyone in the process.
So quick explanation here, I am an computer science engineer, in university I have been studying deep learning and testing some hypothesis, i killed my old laptop by doing so, ai technology is very complicated when it comes to hypothesis with almost about anything you do with it. I did some testing with vqgan and diffusion models when they did come out to answer a personal simple question. Can ai be considered creative. The short answer is no, it is not creative, but please note that it is my testing i might be wrong with approach or execution. So i think you already familiar with diffusion models and how they trained so i won't explain that and straight explain what i did. I trained my model from scratch it didn't know anything, i fed him pictures of very stickmen alike black dogs in a white background with resolution 64x64 and amount of thousands images, i know resolution seems low but it is actually enough to test, and also fed him images of dog's head. And when i requested to cut images of head from dog images it did fairly good job not perfect but it was going in a right direction. And then i fed him images of blue swords and also images of dogs holding a blue swords with a mouth and images of red pistols without dogs. Now what we have model that can recognize dog, dog's head, blue sword, dog holding a blue sword, red pistol. Now moment of truth if i will request him do generate image of dog holding a red pistol with a mouth will it make it or not. And as you might know answer it just give up and were generating mess trying to blend together dog and pistol. So why am i writing all of this, i often hear arguments against and for ai art generation. Some people say that ai is training on data same as people do so it is not stealing, this argument is hard to disprove because nature of ai makes hard to prove or disprove almost anything but we can say for sure that humans are creative and as i tested ai is not, so we already process information in different ways, ai's pattern recognition works in different way than humans so calling out ai for stealing might be true but then again it is not coping it directly, it is using it is knowledge of pattern that just differs from humans. Okey i will finish my observations here and further will just speak my mind about this situation. Ai is perfect representation of our society. Have any of you noticed that every time something original happens like very successful game, movie or some song. Industries and companies will do everything it takes to take piece of this successful cake and they will deny any other ideas whether they are original or not. Most big companies in entertainment industry always make same thing over and over again and the only original works were made are always made by small group of people and when they reach their success Industry giants will start to manufacture idea of this small group in a greater scales to greater profit because it proven to work to profit. This ai situation is exactly the same, if you lost your ability to distinguish ai made art from human made it is not because ai is making original content it is because people are stagnating and making same stuff over and over. Ever wondered why ai so good at generating anime waifus, it is because they are very similar in appearance to each other because we stopped pushing it forward. By making stable diffusion open source we effectively made every human being a corporation, now every one can mass manufacture popular art or art style but nobody will make original art, and even if somebody will make original works it can easily be taken and also mass produced without consent or mention. Ai is here to stay not because it is better in art than humans but because it is better in worst thing about our society today at creating same idea with low effort.
This is very disrepectful and very ignorant to open source ai developers to like Stable Diffusion . Entire concept of ai art actualy comes from understanding how brain detects some specific object in a image.
@@lazarus8453 Ai in the beginning mean to mimic human but every day ai you see now is not, ai word it self that is used today to describe models and other tech is a buzzword that companies use as a marketing tool. All the ai tools that used now do not mimic nor similar to human brain, most of them is variations of interpolation and other math formulas, there is a several limitations that ai can't physically do or achieve and probably never will do to being complex math rather than brain. Of course if they change the core principles of technology itself which I don't thing will happen in near future.
The thing about ai art is that it cant actually "add its own touch" because its not sentient. It finds trends in composition, colours and details and replicates those things based on the works under the keywords imputted by the user. It doesnt have an understanding of why those details are there, it doesnt understand 'people have five fingers' it understands that there are several of these shapes in these places. Ai is a tool made for companies not for artists, and while it can be used to help speed up the thumbnailing process or nail down a composition or colour palette, unless youre an artist that only does one kind of artwork thats all it can do
I've been a freelance graphic designer/illustrator for much of my adult life. I've done a lot of commercial work for smaller companies, and I'm certain that most of that work will be done by AI in the near future. I think it's simply going to be substantially cheaper and faster for a startup or grassroots organization to get professional-looking artwork that way. The issue I keep coming back to is that these companies will get EXACTLY what they want from an AI, no questions asked... and that's not necessarily going to be a good thing for their businesses. I'd argue that the best creative work of this type comes from the push and pull that happens between a business-oriented client, who may not have a very refined artistic or branding sensibility, and an experienced artist/designer who can help make the company's unique vision & identity visually cohesive. Even now, stock photos, generic vector art and design templates are already "good enough" for a lot of small to mid-size companies, so maybe AI just accelerates the trend of homogenized design for all but the biggest companies. Or maybe we'll see some genuinely weird/tacky branding based on the whims of a startup founder after a boozy weekend with Midjourney (that would at least be fun)? Either way, probably much less commercial work for all but the most prominent artists.
I totally agree with you, both with the direction this is going, as well as AI-generated "art" not truly serving its purpose - yet on the surface it seems to do the trick. But it doesn't. I always think of the "art" in an image (or music, etc.) as the way it makes the audience feel. It's all about conveying this particular feeling that you want to convey. AI "art" cannot do that, so it uses randomness instead. Because it doesn't understand "feeling". It intrinsically can't. It was merely trained on images with a description of *what is in* these images, not how the image itself *feels*. This puts AI generated images into this uncanny valley: it looks alright, acceptable, but something is wrong with it.... and we can't quite put the finger on what exactly it is that is wrong... ... yet the executive, who only cares about minimizing expenses, will say: "yep, good enough!" So keep doing what you are doing! They will only get EXACTLY the effect that they want if they go through that creative PUSH AND PULL! :)
its the same thing that happened with video games, there will be little to no innovation by a bunch of rich kids wanting to make a quick billion, which will happen because people are deep in depression and alienation thanks to capitalism, companies are already grey matter and look the same to all of us, less people will question the continued abusive practices and the true citizens (upper class) will not give a damn and push their bland products onto all of us, normalizing AI made "stuff" fixed and catered with an ever diminishing skeleton crew, always happened, always will
In my 10+ years as an artist(3D Animator), all I've realized is that new technologies don't really make our lives easier. As soon as we are able to make things faster, companies just ask us to do more to make up for the time saved. It would be great if AI could do the tedious parts I hate doing, but as things have progressed it's become clear that won't happen. Basically nobody in the AI/ML community has any clue about the intricacies of my work, and doesn't care to. It's easier and cheaper for them to just automate my job entirely and all that will be left is the tedium. Thankfully AI still isn't anywhere close to being powerful enough to meaningfully threaten my work yet, but I hope by that point I hope to be financially secure enough that it won't matter. Also, one thing thing I feel should be pointed out(that other commenters haven't) is that AI/ML can only create something because humans did it first. Meaning that AI/ML "artwork" is inherently derivative in it's current function, therefore limiting the creative variability and leading everything to slowly become more and more homogenized. A problem that already exists, but I feel like AI/ML will make 1000x worse.
Something I haven't heard anyone talk about is saturation. I would think that there will be a huge saturation of art in the next year and we wont know if it is human or ai. Or if a mediocre artists start making hundreds of new art works. Multiply that by millions of artists. Does having billions of new original art pieces devalue art in general?
Great point! I do believe there will be a mass devaluation of art because of AI turning this into a "wholesale" process, but I don't think that will change the "value" of art, it just means art will become accessible and have higher production quality overall
I have also thought about this. But also what about Substantial Similarity copyright. When millions of art or graphic designs are generated every month. And traditional artists would constantly have their art removed because they can't make pictures in hours or minutes and therefore constantly run into copyright claims from AI artists.
I think a further saturation within the art space, where there is a much heavier emphasis on “art as product” and on “art for sale” will fundamentally change the mindset of artists, thereby forcing them to think more like salespeople than like artists that are making art primarily for the sake of making art or therapy and are not leading PRIMARILY from a profit making mindset. This would be a major shift in mindset and in the soul of art. It could unfortunately start resembling the NFT space from back in 2021 and could fundamentally change art for the majority of artists outside of the celebrity art auction houses and for profit markets. Let’s hope not.
Saturation of any product devalues the product no matter what the product is used for. NFTs were supposed to give artists ownership of their digital art. There was already so much theft of digital art even before Ai. I had my art directly stolen. So I helped make a collection that I was very proud of with the hope that it help us digital artists.The scammers moved in and destroyed the whole thing. I can see this already happening to Ai imagery. An artist should have the opportunity to make a living with their art. Unfortunately that does not look likely anymore and the more you work to create a distinct and recognizable style to more likely you are to have scammers steal it. I am sixty years old and I have seen art evolve in many ways. I have adapted and embraced all technology that has come along. Ai is different. It has given scammers the ultimate tool to steal and dilute the art market. This is the first technology I can't get behind. I already see it tearing apart the art world
There is already saturation in art. Everyone's and their dog are posting paintings on instagram. It's impossible anymore to stand out as an illustrator. AI makes it worse, but the problem is already there.
Okay, as an artist, there is a really big misunderstanding that's going around about how artists begin their process to become an artist. I hear a lot of people say that artists simply 'copy' other people's art in order to improve. This may be the case for young artists, aka children, but for a lot of artists in the community we stick to a lot of rules in order to be respectful of others' art. For example, we may copy elements of an artists processes like colour, composition, or linework. These things are simply fundamentals of everyones art that everyone incorporates when making a piece of art. I hear non artists say a lot that we simply copy the whole art in order to learn but this is something that is very frowned apon in the art community and never acceptable. As artists we believe an individuals style is created by external elements like upbringing, culture and also other artists processes. All of these elements are what make another persons art unabled to be copied because it's unique in the way their brain and hand processef that information. In my opinion that's what makes art, art. Please be more careful when talking about the comparison or how an artist vs Ai learn becayse it's not even remotely similar and that's why AI art always looks like a copy of someone's original work. When i look at AI art i can tell directly the artists that were plagiarised to make the 'art' and just because you can't doesn't mean it's original work made from 'learning'.
It doesn't take into account the failed artists. Yes, all artists copy and take influence from other artists as they are forming their OWN style and voice. The artists who fail to create their own style who do not have their own unique voice coming through their work, are not successful. Their work is called a cheap copy, a knockoff. Not everyone who aspires to become an artist becomes one, in fact only a small fraction does. This is where the struggling artist and failed artist tropes come from. Just because someone can copy the work and style of others, it does not mean they are a "good" artist. AI art is similar to the artwork produced by the people who do not have their own voice, who can only copy the work of others, and not be able to convey their own emotions through their work. It is hollow.
@KNYD Yeah, I completely agree that AI art definitely has the look of a failed copy, but in my personal experience, I fully believe artists do not and should not copy for practise. Simply referring to copying the entirety of the piece from the lighting, composition, and linework just everything. I don't like people referring to artists practising by imitating an ASPECT of an artwork as simply 'copying'. Also I think the failed artist trope doesn't really come from people who can't make their own unique look but rather artists that have a unique look but struggle to be noticed or simply from lack of pay that is still a very prevalent issue. But that's just my opinion, it could be different for others, I suppose.
@@KNYD Well, that's major BULLSHIT and/or a lie. Not ALL artists who achieve success are truly original or have their own thing. And not ALL the artists who fail, or simply don0t become mainstream is because they don't find their own style, I WISH It was like you say, but that's BS. If you were right, the majority of people would know who the fuck Neil Gaiman is, and not so much who JK Rowling is, but that's not the case. And I use SPECIFICALLY those two artists, since is well known by real geeks, not modern millennial ones, that "Harry Potter" is basically a COPY PASTE of a 4 number comic story by Neil Gaiman called "The Books of Magic"... But, you didn't know It, did you??
It’s simple the ai cannot think. It’s not thinking or making active choices like artists. To say a machine is the same as human is stupid. We are conscious. We know and are self aware. Look at hands. We understand it’s a hand. We know what it is and how it works. The ai only sees patterns and symbols. It will need images and 3d models to know what a hand is but we know what a hand actually is.
Saying that digital illustrators are "not really creating creating" is pretty uninformed and unfair. Digital painting and illustrating is actually not that different from traditional painting and illustrating, you just have different options and limitations, but the creative process is in many ways the same. And saying that AI is just a tool just like photoshop or digital camera is also very much not true. This is the first "tool" that could potentially replace the artist entirely in majority of cases, thats why it really has a potential to destroy a lot of creativity and a lot of dreams.
The tech and all models must be strictly open source to the public, and companies like Stability AI and Midjourney shouldn’t have the right to profit from the sensitive data of others. When people say AI art is stolen art, it isn’t quite literal but there is a truth to it. It is in fact stolen labour, and there are companies that are using the stolen labour of hardworking artists for their own profits. That is indeed theft.
People need to realize that "quicker" doesn't automatically mean "better". I'm so sick of everyone saying the "old ways" are "obsolete". That's what they said about vinyl records, and physical books, and they're doing better than ever. I don't think the other tools are EVER going to go away. We all have the choice to _not_ use this technology if we don't want to. Tools should be seen as "options", and nothing more. When it comes to this stuff, I don't have that same level of appreciation as with actual art. It's the equivalent of a robot climbing Mount Everest vs a human doing it. It's cool that the robot is scaling it, and accomplishing so much in a quick amount of time, but that's what it's designed to do. When a human does it, it's much more impressive because we're flawed, fragile individuals, and it's so much more satisfying when they actually reach the top. Overall, I'd rather wait for perfection than haphazardly slap on an image and call it a day. The thing about "instant gratification" is that it's there one day, but then it fades almost immediately.
Tbh its just wrong from AI fanatics to say there are no unique art styles anymore and everybody is copying someone, i would consider my artstyle to be very authentic and different because i am only following what my mind comes up with. I am more process driven because it speaks the inner language.
Straight up there’s art styles that are unnatural for some people to mimic because of their pre developed muscles. I did twirling for years and my art is naturally very curvy trying to say draw an original character to fit in in a show that is very triangular isn’t going to work it just wouldn’t it’s going to have a noticeable fingerprint
This went a different direction than I thought. It’s a lot better than many of the other videos, but why ignore legal action or etc. why not show that without the massive amount of data it couldn’t depict images to the same quality. Why did you ignore the huge amounts of differences from human learning and machine learning even as a lampshade would be something. It sounds and uses the language of supporting artists but is strangely in deeply support of AI, basically saying throw our hands up and do nothing but call out bad individuals. Instead of regulation of a dangerous “tool”
I’m inclined to agree with those who have pointed out that your solution would require a large shift away from profit motives in order to work. The problem here is in reality not much different from those we face all over the economy. The issue would be of almost no consequence if it weren’t for the income loss to artists. With a corporation creating these art factories and keeping all the money, we end up with increasing wealth inequality, which is possibly the most consequential aspect of our current onmi crisis. Every artist whose income dries up is a cultural loss to all of society. It may not be one you notice, but add them all together and we get a shrinking pool of human creatives left to bring the truly novel ideas to light. Champions of the tech are too quick to embrace the cynical idea that these machines are the same as people. And like every sector that derives profit from shedding workers, they forget that if everyone is broke they can’t buy their products. If any of these companies want to bring about fully automated luxury communism, be my guest. But from what I can tell they’re quite glad to exacerbate and exploit a growing servant class. And even if that weren’t reprehensible, it isn’t sustainable.
There's a lot of misunderstanding of art in this video,I feel like the biggest one is that an artist doesn't just copy other artists to learn. The difference between artists and AI is that artists can get inspired by anything, not just other art. Many artists don't learn by consciously copying anyone's work. They take things from here and there to build their own thing. It can be a cartoon a photograph, a really cool shaped tree they found on a hike, it can be anything. Comparing the 2 processes like it's 1 to 1 is insane to me. They are nothing alike. Having artists use AI solves a non issue funnily enough, AI art itself solves a non issue, artists make art because they want to first and because they profit second. The art industry is awful to 80% of artists out there, most people cannot make a monthly salary doing art, yet they do it anyway, because they want to.
What I really mind is that art creation is a big part of my growth. I started drawing since I was like 8 years old and at this point is something that I really love, I want to make something I love, something I enjoy. I just don't want to have to stop making my own drawings because "AI makes it better and cheaper"
Every child draws at some point, or colours or otherwise wants to create, it's something I think is hardwired into us, some people grow up and move on, others find different ways to express themselves, and some stick too it for life. I like drawing as a hobby, and I don't plan on stopping because AI is faster or cheaper, and I don't want to give up on human artists ether, if your good at something you like doing, and can get paid for it, that is a win all over, I don't see less money going to individuals, and more to the AI service providers as a ultimately good thing.
Just because ai art can make art doesn't mean I wouldn't pay an artist to create something that I want. I can also get a more exact piece of album art or whatever from the artist. AI is hit or miss. It's cool but it takes a lot of time and prompts to get what you may want.
I am very late, but the AI art issue is about two things. One is the understanding of what is theft and if AI art is stealing, and the second is artists being threatened by a more efficient method of producing artwork. Something lots of people don't understand is that AI art doesn't use stored image from training data, it uses parameters based on noise maps made from images, which look like static. The training data is terabytes of size, while the AI itself is only gigabytes in size. Their artwork is not being sold for profit, as the images made are so far removed that they can't be considered identical, and are all transformative. This is why copyright lawsuits against AI art generators have been failing consistently. Artists are threatened by AI art because it can make products of immensely high quality for practically free. Artists will maintain a niche, and some have even benefited by raising prices as a response to AI art, but now art creation as a whole is accessible to the masses, rather than locked behind a paywall that feeds into artists' incomes, or behind a massive time investment to become skilled at it. Also, the point of "unfair capitalization" seems odd, as there seems to be nothing unfair about developing a better product to accomplish a task. Otherwise it would be "unfair capitalization" to use excavators, bulldozers, and cranes in construction, rather than 100% human labor. Also, you can't claim a style, it's simply not possible legally, an arbitrary line anyways. By claiming styles you would bar basically anyone from drawing in various styles, preventing people from making comic style and manga style art as an independent creator, or turning "How To Draw" art books into legal traps. It is an unironically tyrannical concept which would be an absolute disaster for the art world. Also, 19:33 that's what she said lmao.
I think the issue with being able to copyright art styles is that, it then means you will have a lot of problems in terms of other artists suing other artists because their style is too similar to their own. And how on earth would that be measured? How can you look at it and say if it's too similar or not? What would the determining factors be? This is likely the reason an art style can't be copyrighted as of now. In my opinion, the way it would work best for both parties is if artists simply get paid for their work. Let's say an artist knows their art has been included in the data set (though I do believe it would be better to be opt in). They could be able to go to the AI tool, and make a claim and with validation that they are indeed the copyright holder, they would then get compensated for every generation that used their work. Basically, they would get a certain percentage of whatever income the tool makes. You know similar to how Spotify works for music.
I mean, yeah, that would be better than how it is now, but right now there's already an issue with AI not being able to unlearn what it analized from an image, so it still gains something from it even if it is deleted. The otger issue is that spotify SUCKS for artists, they pay the least out of all the music streaming apps with maybe better pay per stream for big/mainstream musitions, the thing is, it's kinda functioning as a monopoly. I do wonder, however, what if there will be a new category of enforcement of creator's rights: artists could copy an artstyle BUT it can only be enforced against AI genarated images.
@@_loss_ Not entirely sure what you are trying to say. Putting something out to show doesn't mean it doesn't have copyright protection. So for example, if you see an image, sure you can view it, and download it. Nothing wrong with that. But if you sell it, or use it in another project and sell that work... then depending on circumstances that can be considered copyright infringement. The question really is, does using copyrighted works to train an AI for image generation count as fair use. We will not know till a court case occurs and the court decides.
An Ai can measure how much and what styles it uses and tell you, but it doesn't and AI users probably won't show what prompts they use if they are using someone else as reference. How come an artist can show their timelapses and references and an AI user won't be as transparent?
As an artist, I do believe art to be process driven, and yet, I don't believe that AI art lacks a creation process. AI democratizes the creation process of art and gives that creation process to people who, for whatever reason haven't or even may not be able to go through the steps of traditional training. Not everyone has the mental or physical capacity for that kind of training or time to work towards it. Who are we to gatekeep methods of creating visual expression through text based generated AI images? How can we say, your imaginative creative expression of yourself facilitated by technology is not art?
The issue is that AI won’t displace artists careers, such as Concept artists if the databases that the AI models like Stable Diffusion are built with didn’t steal millions upon millions of images under the guise of research purposes that were then conveniently used commercially.
Yep! Similar AI models in the music industry are trained on copyright free works.. because otherwise they'd be sued. They have issues with overfitting meaning that sometimes you can find a whole chunk of someone's song in an AI song.. THE SAME issue that the AI art has. Look at half assed signatures in the corner that the AI tries to replicate, for example. The law industry is a joke- supposedly to protect everyone's rights, but in reality, only if you have the money to jump from court to court as a multimillion company tries to bankrupt you to give up. If you make a collage that you then sell, you can't use copyrighted images for that. Even for photobashing there's bundes of photographs you can buy to legally use. When you use copyrighted work to make a machine that you will sell licences for and the users will sell artworks from.. same applies, imo. Sure, the art is "not in the finished art". If I stole a truck to build my house it'd still be llegal even though the house doesn't contain the stolen truck.
@@josehumdinger6872 Bro, he's not making money off of it and if he did, he would be sued - can't say that about the AI database which is obviously also fed with IP-data if it can react to a prompt referring to an artist. You win dumbest comment of the day
In the cover of your video, it reveals two version of your portraits. How much it cost? AI companies used artists works included some of my work which was in Laion 5B to train a commercial use AI model. They didn’t reproduce or transform any of the files in it but directly feed those copyrighted artwork without consent. Then, the machine “deep learning “ it. How do people define its “learning?” This word is only that scientists created to let people understand how it function easier and trying to say that! “Hey! Machines can mimic what human do!” I’m an award-winning illustrator and spent years to develop my style, but AI companies scrape my art to train the AI without my consent. I can actually create a unique portrait, but now, will people pay me instead? A digital quick portrait is minimum 200 USD for only personal use. Every generated image for profit use means one human artist with that style short the money to keep them surviving in this unethical world of AI. Accordingly, it is not a tool for artists now, and AI companies are did it wrong from the beginning. Those people are still build the lies to make people hug and accept this unethical tech…
Yeah, it's mostly nonsense to not see this as a blatant violation. It's just impossible to have the vast database without doing unethical stuff and furthermore as you correctly point out, the idiotic reductionist claim that a machine "learns" and the equivalency made between a human process and a machine process is mind-numbingly evil. It's sad to hear about your situation. I switched to motion design because the illustration/art market seemed so crazy but I know those mega-corporations psychopaths are gunning for me too. I'm thinking of switching careers, let those fu** rot and changing to physical art (painting and sculpture) without the aim of commercialization. I just don't trust people in general, they all talk lovey dovey but if money is involved most won't value art and craftsmanship.
Excellent video! Best explanation of AI Art ethics problem. The magic paint brush was a hilarious prop. I thought the video was very informative and not too long.
Personally, as an indie gamedev and game design student, its gread for creating general moods or ideas and its especially Usefull for fast-paced prototyping. In the end if your making the final product you will still need artists. The artworks created this way never see the light of day. So i feel its great for getting inspiration, same as going online and looking at art pieces for inspiration, but i can controll more factors.This is a very difficult complex topic, and we should add to this discussion. Its not black and white, but AI art defenetly has its risks.
Your videos are next level! I had never thought of art as a process before today. I’m not typically a fan of longer videos, but this was well worth the watch.
Digital illustration still has a process behind it. It’s still art. It’s still creating. It’s not any less valuable because it’s not traditional art. Insinuating such is honestly really offensive.
The problem with art appeared after WW1, when dadaism claimed that any random event is art. Well computers can do more beautiful random events. But what is missing in computer generated art and in dadaism, is intent. An artist made an art to show us something, the generated art is something that is happening without intent, is like blooming of flowers and drift of clouds. Only when the flower is drawn is becoming an art object.
So while I agree that in general AI is here to stay, I do not agree that this particular use case, application, and business model is here to stay. When you speak of the fact that it's an impossible problem to solve to try to credit or compensate each copyright holder in a dataset, that is tuff titties for the developers, not the general public. "We'd love to NOT steal your work, but it's just too hard and our technology MUST move forward" No, I'm not buying that. Here's some, IMO totally reasonable ways this can move forward: 1: AI can use exclusively images in the creative commons, and artists who opt in. You can create a myriad of incentive models to get artists to opt in, that's a huge discussion, but that's the main idea. Almost certainly, this would result in the model not being as good. The model not being as good doesn't prevent the tech moving forward. It ONLY stops it from freely exploiting the property of people. 2: At any point in time, artist should be able to blacklist their names from being invoked in image generators. I can't think of ANY legit reason to not have this be the case other than companies specifically wanting to exploit and steal from artists. I am not a genius. Someone smart has thought of this before. It's NOT being done for a reason. It is MEANT TO STEAL. A few additional points - This fight is currently about art, but everyone should be in favor of some common sense protections about the usage of data for AI in general. Right now this might seem benign to the average person who all in all doesn't care that much that this is stealing from artists. But understand that everyone has images of themselves all over the internet as well all over social media. Lots of it conveniently tagged with who is in the photo. What is all of this was put into a model and trained to be able to invoke your face by name? If anyone could create images of you, your children, your family, doing literally anything. If anyone could take your face and make and sell merch with it on it, then tell you they don't owe you anything, that your face was freely available on the internet, that it would be too hard to NOT use your face. Now consider that for many artists, their art and their style is more central to their identity than even their face. There are many famous artists out there that hardly anyone knows what they look like, but they could recognize their work a mile away. It's not just someone's "product" or even their property being stolen, it's their whole identity.
>If anyone could create images of you, your children, your family, doing literally anything. If anyone could take your face and make and sell merch with it on it, then tell you they don't owe you anything, that your face was freely available on the internet, that it would be too hard to NOT use your face. Now consider that for many artists, their art and their style is more central to their identity than even their face You are mixing up one's right to their image, their brand, which is a long established right - you can't just create a fake 3D Tom Hanks and put into your movie and say it is Tom Hanks. You are mixing that with the right of one owning an art style, which do not exist: anyone can draw like you, art style aren't copyrightable. They never were (Disney would love that art styles were copyrightable btw). So even in a world where artists "got their way", and you couldn't use other people image without "their consent" to train a database, there would be nothing preventing companies from just hiring some skilled artists to copy your style and than pay a few bucks a single time for having the right to THEIR paintings, and use them to train the model. Also, this world where "anyone can create any image" pretty much already exist. Anyone can go to your facebook, or instagram, and download 10 photos and use Stable Diffusion and Dream Booth, fake any image they want. This world already existed even before that, with the invention of Photoshop, it is just easier now. It is all open-source and freely available, the genie is out of the bottle.
@@someone1861 I'm not mixing these things up. I'm simply making an analogy. In any regard the artist doesn't need to own a style. They own THEIR WORK. That's what's being used. It's not "their style" that's being used, it's their actual work. The generators interpretation of their style is the output, that's a product created. A product created by directly using something that the company doesn't own. Each person has the right to not have something they own be used to create someone else's for profit product, or at least they should.
@@midlet Copyright isn't actual ownership though, it's a limited monopoly over reproduction rights and substantially-similar works. "IP" are not equivalent to physical property (and there's even a strong case to be made that the legal construct called "IP" infringes on physical property rights!) I could also see some form of dystopic logic being applied here in this thought experiment: If you listen to music while you make art for a client, the music is influencing your mood, your thinking, and thus your art. Does that constitute "using something you don't own" to create something else for profit? I would hope that most people would see this as a ridiculous standard to hold, but there is probably some executive in a record company who dreams of such scenarios.
Your "AI Art Proposal" would unfortunately have serious negative consequences, despite your best intentions. The main purpose of copyright law is to promote the creation of new, original works. By restricting the use of specific styles or techniques, copyright law would inhibit creativity and innovation rather than promoting it. Many artists learn and grow by emulating the styles of others, and some styles become whole movements that define an era. Additionally, art styles are often abstract, fluid, and overlapping. They evolve over time and can be influenced by various cultural, historical, and individual factors. Defining an art style concretely enough to be subject to copyright would be extremely challenging, if not impossible. And finally, copyrighting art styles would lead to the monopolization of certain artistic expressions, which would be against the public interest. It would limit the variety of artworks available to the public and potentially stifle new creations.
The problem is: why are so-called "Artists" an exception? AI feeds from everyone: from actual photos taken by photographers, to generate photorealistic art. From code written by programmers, to generate code. From articles written by writers, to generate text. From scripts written by screenwriters, to generate new scripts. And so on. This will hit everyone and NOBODY is exempted from this.
They’re not, you probably only hear from them the loudest because of what’s fed to you by the algorithm. Everyone you listed is on the exact same side and is having the same concerns and problems. It’s not artists begging to be exempted, it’s creatives and workers asking to not be exploited and profited off of without their consent or simply disposed of.
This doesn't take into account the failed artists. Yes, all artists copy and take influence from other artists as they are forming their OWN style and voice. The artists who fail to create their own style who do not have their own unique voice coming through their work, are not successful. Their work is called a cheap copy, a knockoff. Not everyone who aspires to become an artist becomes one, in fact only a small fraction does. This is where the struggling artist and failed artist tropes come from. Just because someone can copy the work and style of others, it does not mean they are a "good" artist. AI art is similar to the artwork produced by the people who do not have their own voice, who can only copy the work of others, and not be able to convey their own emotions through their work. It is hollow.
I don't think AI art is an apocalypse like so many people make it out to be, since people have show time and time again that people like Human made things. You can make incredible music without hiring a single musician and have been able to for well over a decade, yet musicians still exist because people like human played music even if it is possible to recreate it entirely with computer programs. Physical calendars are still selling well despite digital calendars being a thing for ages. etc, etc. People like human made physical things. Wait a decade or more and you'll find that AI art was not the apocalypse so many made it out to be, Human made art will always be popular.
as an artist, I feel like you were missing the point that most other artists have with ai. its not the money making element, bc that is inevitable as you put it. its that these ai are being trained by non public domain images in the first place. every image on the internet belongs to somebody, and the people who made them should have been asked if they wanted to opt in or out before the ai was trained. then you have a complete list of who's art is in it. if im not mistaken, sometimes ai will spit out an image that's a direct copy of someone's style, even if the person didn't put in a prompt asking for it. but only the artist or someone who knows their style well would be able to tell, and the odds of that happening are slim to none. this whole video I was waiting for consent to be brought into the conversation but it just never was
> *missing the point* TBF there isn't some singular "point," many reasons, or combinations of reasons depending on who you ask. > *by non public domain images in the first place.* IMO that is kind of an incorrect pointing of the problem - since in a few countries, including the US, anything eligible for copyright is copyrighted, and that includes works you volunteer or allow, and creative commons works. It confuses licensing status (and whether licensing is needed) with copyright status as if they are synonyms, when they are not.
I think your take on the ethics of the AI I think miss a couple things. First, the artists who published their artwork online do have a license associated with it. If that art is then used to feed the training of the AI in violation of the license, then there is a very real legal problem with the people who chose the training data. For example, if art is just public domain, then there is no problem. However, creative commons style licenses often have attribution clauses and/or non-commercial use licenses. The generated AI art doesn't have the necessary attributions. The big challenge here is a legal one. What constitutes a forgery now has to be refined. It's not the first time that new technology has forced society to adapt. _Some_ of the same arguments were applied to photography. The advent of AI art, which is a very new capability which is going to force our society to adapt. It's an open conversation, and we as humans are still trying to come to grips with what should be the ethics of training AI/ML, and the implications of what that new capability provides.
all countries work differently, sure you have copyrights to your artwork in most countries, but also art can be created from other artworks as there are many such cases. It just has to be different enough.
I think the only way for this to work and it not screw artists over (because some of the options such as oh you should have something like copyright over your creativity could cause things like people having to pay photoshop when they use their brushes) is just the requirement of consent. Consent for every image the thing is trained on, for that consent not to extend to other ai and for no ai to be comprised of more than 15% of one artist’s work (or that if it is they are given partial credit but would have to agree to every image used) and I can hear people saying’but it’d take weeks or months to do that!’ Yeah but months is shorter than the years or decades it takes to refine your art from scribble to masterpieces
Artists can have whatever opinion they want about what defines art and what makes it valuable to them. But as soon as you bring in a consumer the artist's opinion on the subject isn't the only one that matters. Consumers get to choose what is valuable to them, and what they want to spend their time and money on. And this can't be an either or where AI art isn't allowed to exist simply because it brings competition and possible loss of income to traditional artists. If that's how society worked we wouldn't have cars because we have to protect the livelihoods of horse breeders and buggy makers. AI art gets to exist and traditional artist will have to adapt to having that as a fact. The only part of the artist's argument I can get behind is AI directly using IP of other artists to create without giving credit or compensation. I say all this having been an artist all my life.
Excellent breakdown and suggestion for how AI could be used ethically. I know it would be immensely profitable if a famous artist with a distinct style (Sakimichan or Artgerm for example) sold an AI engine solely trained on their entire vast body of work, and clients can pay to use it to generate quicker, smaller commission work. The AI would also automatically print a watermark or signature on the final piece after processing so people know who it was sourced from. And both sides benefit too - the artist owns their style and monetises it, and the client is able to purchase access to get something in a style they love while supporting the artist.
*Context of my comment:* I am somewhat of a hobby artist, I like drawing and my family and friends keep telling me I'm not at all bad at it, even though I myself see many flaws in what I create and that I need to learn and improve. *My opinion:* I see art as product driven. 'Process driven', as you clearly show in your explanations, is just another term for 'money driven'. Furthermore, I can understand a true process driven argument in this discussion. But even then I argue that AI also goes through a process, just a little faster and perhaps a bit different process than the process a human uses to create art as biology (human) ain't no binary logic (computer). However, is the fact that an AI is faster than a human a reason to disqualify it as true art? If so, does that mean that the work of an experienced artist, who can create faster than a novice, is less 'art' as well? That's just ridiculous
Most interesting commentary made here, very good insight and conclusions made about this controversy. In my opinion, from past experience where I personally went through 4 years of commercial art- illustration career training back in the late 1960's and early 1970's using traditional media for creation, technological change has created new opportunities by destroying older forms. I recall back in the 1960’s and 1970’s how illustrations and advertising art were created using tools that are today to a large degree obsolete. These older tools took much more time learning and practice to use effectively. Such things like rub on graphic letters, color separation on transparent mylar type sheets, and other tools for print ready products. The use of computers and software made most of these old techniques and job skills obsolete. The problem with this change is that the human mind and ability to keep up is not always there. There are some things that one can easily transition to and from yet others that are not within an individual’s abilities to migrate. We all have things we are good at and not too good at. You need a lot of luck, kind of like beating the odds at winning the lottery to keep a stream of success continuing through technological change that displaces long learned valuable skills. Looking at past precedent and historical evidence throughout time, most generations lived in an era where a set of job skills would last a lifetime and were valuable for survival and making a living, change came slow and more easily comprehended. The present generation on the other hand has seen too many changes in one lifetime that has totally disrupted valuable job skills to make a living and put humans into the shackles of constantly spending valuable time and money keeping up with the rat race of accelerated change. Hard earned learning skills and education have a much shorter life cycle of use in today's fast changing technological pace of moving forward. This pace will continue to accelerate as a necessary means of survival as the world's population continues to grow exponentially and a means of managing the growth is crucial. In all fairness a lot of changes do provide convenience and make things easier yet at the same time de-values any worth for some time, effort and dedication towards the acquiring of any skills. Honestly, has all this technological change really improved our lives in a measurable degree that makes all the efforts to keep up with the rat race worthwhile? Bottom line being a worthwhile investment. Would the other possible outcome be more in line with the reality of what all this change is doing to our human race? Are we simply making things more complicated and mentally fatiguing as it destroys our ability to make a living and forces one to constantly learn new skills that are like rolling the dice? Is there a possible connection with how all this technological change is destroying human morale and leading to more hopelessness and loss of confidence in one's ability to make a living from personally developed skills that are valuable in the marketplace? Seems the rate of worldwide protest in opposition to different things is growing increasingly in this generation along with an increase in crime and corruption within most countries and people roaming around the world being displaced out of desperation. In all due fairness AI was not the biggest competition and challenge to creating art from scratch using traditional mediums or digital painting and drawing. The same thing happened over a century ago when photography became available to the consumer. Before photography many artists found work in portraiture since that was the only way to create a likeness of any visualization of a human as a means of memorabilia. Likewise, yet to a lesser degree landscape paintings, and still life. Photography won a large percentage in competition by replacing traditional medium art of drawing and painting for freelance and other established business means of making a living.
14:57 If you think "Unfair capitalization (basically, exploitation)" is the problem, the real problem is actually capitalism. Exploitation isn't a bug it's a feature
Thanks for this balanced video that expands on the valid points of the artists as well as the developers. 6:18 is a nice summary of the whole situation! I've actually seen both sides of the arguments _within_ the artist community. Paradoxically, it's usually the _product-driven_ artists that were the most upset. The value of art as a process is a also an emotional one, and ultimately what drives an artist forward. The product is the destination and does also have layers of value and meaning beyond just its monetary aspect. But that actually goes for AI-generated art and the atmosphere it creates by capturing the unique nature of a particular style. There are few words that have such a contested definition as the word "art". ("Love", perhaps) I can see why there is such a huge debate about whether "AI Art" meets some of these definitions, like an author's self-expression that's not given by something that's generated by an algorithm. Yet there is also a common notion that there is no actually original art, and what differentiates plagiarism from professional art is knowing how to steal from many different sources. In fact, some of the same artists that now talk about AI just stealing art argued about precisely that. (As the book titled "Steal Like an Artist" suggests, which I actually found through a TH-cam artist recommendation) And as you mentioned at 10:15, plagiarism is about _the person_ falsely claiming work as their own. Whether that be an actual artist, or - In the case of AI art - the owners behind the technology. Don't demonize the tool - blame the user. In that regard, AI basically taps into more sources of artworks as reference to create an images than any real person ever could. Of course there is a more unique conversion and perception when an actual person converts their reference material. But the technology is not actually taking any one artwork and selling it as theirs. That said, using the images without consent _is_ an issue, as you pointed out at 9:05. Thankfully, opt-out options are at least popping up that allow artists from excluding their art. But while Opt-out could still be considered a crummy approach (and laws in the future might even make this outright illegal) that ultimately won't prevent AI art to reach a level where it can meet the vast majority of art commissions. I think the real issue is the underlying fear through the sudden shift in the art market, as noted by you at 11:03. Suddenly, technology made it possible to create art styles that previously required hours of work - let alone years of practice to hone that skill. So I can see why full-time artists are so upset right now. Particularly the less professional ones (as there is no target group left for them, losing to both cheap and clean-looking AI art that the average joe is content with, and not being able to keep up with the pros providing refined and expensive hand-crafted art for the connoisseurs.) They now meet the fate that many jobs were previously done manually and lost to automation. Art was considered this bastion that computers and automation were for the longest time impossible to touch, but now suddenly reached a dumbfounding level of proficiency in the blink of an eye. Making a living as an artist now definitely got harder. But in magical wonderland where money is not a issue and merely becoming a good artist and enjoying the process is the goal, the situation actually got a lot better. Because ironically, It's never been easier to create reference material to practice drawing specific objects, compositions, drawing styles or color palettes than before. But as you implied at 11:16, this is a blessing for the society at large by making art products far more accessible than ever before. For the consumer. For the hobbyist painter. And even for the _nonvisual artist_ that previously had no budget for enhancing their work of writing or music with pictures, this is a win/win situation.
Thank you for watching everything and giving such detailed responses to some of my points! I'm glad this has spurred more discussions like these. I've also been recommended "Steal Like An Artist" and as an artist, I actually believe the word "stealing" in art has got a lot more negative connotation than it should have. As humans, we learn by imitating and copying and there should be no difference between AI doing that and us doing that.
@@cynzyyy Thank you for providing the basis for these discussions! You think "stealing art" has a more negative connotation within the art community? So far my impression has been the opposite: That the idea of "good stealing" (as opposted to plagiarism) is widespread in the artist community, mostly regarding "stealing" as a good habit or necessity even; for precisely the reasons you mentioned. Whereas the average consumer is much more likely to look down on taking elements of other people's art, often even due to the misconception that a 'talented' person can simply draw from memory, and using references is cheating. Either way, it probably explains why the louder parts of social media largely speak out against AI.
I think this is a similar disruption to what record players were to musicians around the 30s. It doesn't fully replace live performance, but there are kinds of demand that can be satisfied with this mass produced art-like thing. Of course, for someone who values the process will not be satisfied with an AI-generated picture. For the rest, AI will outcompete artists by being cheaper. I don't believe it's plagiarizing to AI generate "art" in the style of a human artist. You could just as well pay an artist to copy the style of another artist, and it's still a separate creation. The only difference is that now it's cheaper. Of course it's nice to specify if one takes a lot of inspiration from someone else, but it's not mandatory, because it's a new creation. Whether it's for profit or not shouldn't matter. Copyright protects the concrete, materialized creation, not the style.
6:20 Answering your question about how I view art: Is it fair to say I see it both ways? Both as a process and a product. My thought at this point is to mention that the law seems to agree with this view. AI "art" is a product, but it can not have copyright. Meaning it's a product that lacks the process. It can exist, but it isn't protected.But this is where, I think, the discussion takes a turn, because the artists, who's work is protected by copyright, says "we want to exclude our art from being a part of Ai training", and is that even reasonable? Copyright includes language about transformation because if a human artist uses another artist's work as inspiration, we accept that. But as soon as we turn that human experience into a process to produce a product, we want to be able to control it. And, this is what _I_ think is what's at the core of the discussion. It's emotional. We think art, creativity, and the ability to create is a uniquely human thing. Then along comes a machine we've make something that turns that human experience into a text prompt with an output that matches anything a human could create. If that's possible, what does that mean we are as humans? What do we have left? Okay, gonna watch the rest of the video now.
I think your solution to allow artists to only opt in is either short sighted or masking a desire to kill AI art. You repeatedly site an example where the prompt is "This in the style of a specific artists", but a lot of AI art prompts are more broad than that. "This in the style of a digital painting" or "Photo realistic" or "children's illustration", and in those cases it's still not a signle artist who's work is being is being infringed on, but everyone who's art can be described in that way. Even in your example of a singular artist being parroted and their ability to opt into that, the most prolific artist in the world hasn't made enough work to begin to train an AI. If a single artist never published, say, a bowl of fruit, or an anime inspired mecha, then we can't use their dataset to make a Carmen Maranda mecha, because the dataset doesn't include examples the elements that would go into it. It's not just working with one hand tied behind it's back, it's working with both hands tied, a blind fold, and in a language you don't speak. Imagine we decide that we don't want human artists to copy each other. Well, if you decide you want to be an artist, we'd first have to figure out if you've already seen anyone's art every, at any time. If so, you can't be an artists. But now we have a problem that no one can be an artist. However, since artists are obviously important to a thriving society, so we decide that we're going to take a certain portion of the population, kidnap them from their parents, raise them in rooms with plain white walls, and hope that they'd magically become the next Monet. Oh, except Monet trained under and copied artists of his day, so that's never going to work. At best we're going to send ourselves back to cave paintings. Cave paintings of rooms with white walls. I hope my hyperbolic example makes my point. Trying to control and limit the training data doesn't simply limit what's possible. It makes it impossible. And if that's what you want, then just say that. Be honest. You don't like AI art and you want it to go away. That's fine, if that's the position you want to take. As I said before, I feel like AI art threatens what it means to be human. We're taking something that we thought was uniquely human and this inhuman machine is producing it with comparable quality. That scares us to our core. But there is hope. Because in reality there's nothing that a machine learning process can do that hasn't already been done. Sure, it can remix in ways that looks like creativity, and in a lot of ways remixing _is_ creativity, but it can't create anything new. The frontiers of imagination are still the domain of humans. We've seen this before. When photography developed to the point of threatening portrait artists jobs, yes, some artists lost their jobs, but other discovered new modes of expression. Cubism, surrealism, abstract, art deco, and a whole host of new artists expressions were invented. Meanwhile, people with cameras were perfecting their craft with amazing framing, double exposure, and other tricks. This thing that we though would destroy art actually expanded our ability to create. I think AI will do the same. It's not a destruction of the process, it's the evolution of it. Unfortunately this won't do much to satisfy the emotional reaction of those who feel personally threatened or slighted by machine learning, and I get it. I guess I would recast your argument as not a fight between Product and Process and a fight between Reason and Emotion. The Emotion says "You're stealing our process" and the Reason says "It's just a Product". But more than that Emotion will pound on anything they can to get attention, and Reason, I'm afraid, is gonna just ignore them and move on.
This video kind of gets to the problem but it doesn't address it: It is possible to make a fair system to create the databases from which the algorithms will learn. AI companies are processing artist's work in an unfair and even illegal manner, and these companies need to be held accountable for this infringements. An not perfect example on how to do it correctly is how music generator algorithms treats musical copyrighted material
Great efforts in contributing into the discourse. I think one bit that's lacking in the whole discourse of AI/Art/Design/Visual creation is notion of monetising "authenticity" instead of "time"/"emotion of the artist". What makes one type of art more authentic than the other? AI allows us to shift towards producing ideas faster. Instead of the reliance on craft/"process" driven. I think it'll also be very interesting to bring in perspectives of generative artists.
Also, that's not how copyright works. It's not plagiarism if you create something in someones style - a style can not be copyrighted, only individual works. The copyright infringement claim by AI is mostly about the files that it generates and stores from the images it analyzed.
I agree with almost everything you said, but I do think that the onus should be on these companies to track the images they use. The datasets should *all* be publicly available, the artists should be able to find their artworks in these sets, and they should be able to opt-in/out. There needs to be more responsibility on the part not only of the companies providing the AI art producers, but the people who put together these datasets. Right now, it's all "I don't know, I just used it, I just found the image and included it" etc. But that's really not good enough for something that will cause such a profound change in society, in people's individual lives and livelihoods, in the accessibility/valuation of art. These people and companies need to get themselves in gear - they are mostly individuals or groups of computer nerds, not ready to deal with these kinds of changes that they have brought about. I respect that, and I also respect that it will take a lot of time and money to implement my idea. But I think that's more important. AI is supposed to be efficient, but it should ultimately be a *choice*. I should be able to choose whether I go to a human artist or computer. I should be able to choose whether I want my art/photo/likeness included in a dataset. I should not be told that it's too much time and money and, therefore, that that choice is a waste of time and money. These are things that should've been thought about long before this point, ideally. But they weren't. So we need to do it now, while it's happening. And if it messes up the AI revolution for a few years, I think that's acceptable for humans to be properly respected AND responsible.
Alright why would 24/7 guy care that artist gets replaced? He dosent really care that it is made by a Human or a machine. If he likes it, he will buy it.
It is way more easier just simply to watch the definition of art, and everything will be clear: "the expression or application of human creative skill and imagination, typically in a visual form such as painting or sculpture, producing works to be appreciated primarily for their beauty or emotional power." - Oxford Languages
I don't normally comment on YT but I just found your channel today and I've been binging all your vids! Thank you for such beautifully presented and thoughtful content. As a concept artist working in AAA games for my work, the ethics of AI really does concern me and I do think we should be questioning the morality behind it and where to draw the line. I really appreciate your research and insight into this.
thats wild!!!!! I was just thinking about how works are/content is created, before watching this video, and settled on the mantra "the product is the process, and the process is the product." I wonder if this conception of human-created works will be discussed later in the video
It's not just a tool, it's a replacement, let's not just talking about business and all that, using AI to create something for urself, maybe just hanging on ur wall or just making it for personal use wallpaper was also a threat to any artist out there, it is stealing too doesn't matter is it for personal use or commercial use, only the artist themselves were allowed to use their own hard work to train a AI model for themselves, not anybody else that was not given permission to people was actually commissioning AI to create art, not 'using' it let me repeat, it's never just a tool, but a replacement, u can already tell by how many lesser famous artist's career was negatively impacted by it, there were plenty of tutorial out there teaching u how to use the 'lesser famous' artist's works to earn money, just one or two stealing jobs from that lesser-known artist was enough to kill his whole career
AI art will not kill artists. But it will replace mediocre artists for all the common work (minor graphic design, industrial drawings, cheap commercials/logo etc..;) Artists will still exist but there will be no room for "average" artists, they won't have a job opportunity any longer. Just like how industrialisation displaced artisans work. Artisans still exist, but only for high end products.
20 minutes I my ideal length of essay video. More is welcome. This is the best articulation of the nuance of AI art. I have adopted this perspective in it's entirety, it's just really good.
Thank you, Cynthia! Another great video! Here is my take on this: In the end, the questions are: Who are we? What do we value? What makes us human? How are we different from non-humans, from machines, or even from each other? What makes us unique, both as humans and as individuals? My understanding is that whatever we are, what humans want is to connect to other humans, in various ways, art being one of them. (Seen with Darwin’s eyes, ultimately we connect as a form of expressing our phenotype to find someone to mate with, and a community that helps ensure our survival and the survival of offspring. Our brains are set up to want and do this. A tangent point is that we could of course share art with animals, as part of our community, so it is not just about humans. But all of this only sets the stage for what living beings do, so let’s move on to see what happens on this stage.) So, art is by necessity a form of human expression. We care about the people who create the art, about the provenance of art, and what thoughts went into creating art. - Even if we have to find our own interpretations and use art as an object onto which we are projecting our own thoughts, it is still about our human interpretations and projections, and how they relate to the art and its creators. An interesting thought is: if an AI would not only create art by closely following specific human direction, but if it would create many pieces of art by following an automated process, would we still be interested in every one of the many pieces of art it “creates”? - My answer is: no, we wouldn’t. Because all this “art” that the automated AI generates would just be the product of some process within the automated AI. And unless this “art” connects us to something of value to us, which ultimately are other humans, life, community, we have no reason to care. A slightly different question: what if someone selects “interesting” pieces of “art” from the stream of creations of that automated AI? Would we be interested in those? - My answer is: yes! Because in this scenario, there is a selection process going on, and the human who selects the pieces of art and the reasons why they selected them are interesting to us again. The selected pieces of art connect us to another human, life, a community. The AI is only a tool used in the process of this creation. Now, as we value human-made art, even if AI is used in its creation, we also want to value the human creator, and their work. And we recognize rights that come with the creation of art, such as copyright. So, in the end, the core problem with art that has been created with the help of AI is that it makes the imitation of previously created art so much easier. We need to see how well copyright and its tests can handle this. The usual questions are: Is it a copy or a transformative work? And by how much? - And how do we handle the mass-creation of imitative works reasonably? Is the Millennium Copyright Act sufficient? What can we learn from similar problems on platforms like youtube? I do believe that AI-generated art gives this problem a new dimension, because we probably need to define the boundaries between a copy, an imitation and a transformative work better. And what is “fair use” in this scenario? And we need to learn how to deal with these problems at a much bigger scale. But fundamentally, we do have a mental framework that we can apply to deal with AI-generated art. We just need to learn how to apply this framework to the new challenges. Oh, and how do we do that? Ok, I believe we need platforms that help with copyright tracking, consent arbitration and payments. So, what youtube does, just better. And partially automated, but with ways for humans to take part in every step, such as discovery, arbitration, negotiation, etc. And btw., another form of AI can help automate this.
what someone said that stuck withme about this is that in the 1960s, they thought machines would free us from drudgery and let us do things we enjoy (ignoring the issue of companies just choosing, or being forced by economics, to simply discard workforces, and the difficulty of retraining), now the machines are here, they write code and draw pictures and write papers. Great, so if machines fill in for the drugery, and machines fill in for the art, what's left? I really think this is a situation in which the genie is out of the bottle and we're going to be feeling the bullshit coming off chatGPT for years to come.
I think something to acknowledge about "Misunderstanding 1" is that it's not just the user being a 'garbage human being.' They're just using the technology and if they ask it to make a portrait of themselves in the style of X artist and then want to use that on their social media - that's murky territory. The tool itself needs to have safeguards in place, and that falls on the companies making these tools to not just rush things out for hype, but to actually consider how they're implementing this tech. A simple example is GoogleMLM (a text to music AI tool), if you say, "make me a rock guitar lead line in the style of AC DC," it WILL NOT do it. If you use your own words to try to explain to GoogleMLM Angus Young's general guitar style (not an easy thing to do) then it will. This is a really basic level of protection for existing artists who are concerned about how their artwork is contributing to these tools, potentially to the detriment of their livelihood. To be clear I'm hugely excited about AI in the art space. As a very mediocre musician, having AI tools to help me with the songwriting process, as well as mixing and mastering is huge.
Also for "Misunderstanding 3" the AI is not adding new experiences, it's just regurgitating what its seen based on user inputs. And regarding your question of 'is there a solution?' the AI is trained on a set of data. Companies should be responsible for the data they put in and it would not be unfeasible to create a sort of royalty structure similar to how music royalties (esp for streaming) work today. Lastly, your proposal couldn't work, right? I don't think most artists would want to outsource their work to AI, and also their own work probably wouldn't be a big enough dataset to run an AI image generator off of. Again, I think AI is great, but when your gut tells you that for profit companies taking copyrighted artwork to train computers on, and individuals developing their own style to go on to make their own art and sell for profit... it just feels different.
If given a choice, most artists would rather spend 20+ hours on a piece than have AI generate it [Unless that artist wants maximum profit or can make the same living generating images]. Otherwise AI art already has a stigma attached to it. There are many tools that allow artists to be more efficient and accurate. Like the camera lucida and camera obscura. But - because somewhere in the 1900s someone labeled these as "cheating" no no one uses them. "Leonardo would never have used a camera obscura. Leonardo is amazing and Obscura is cheating and cheating bad! Leonardo good!" - even though Leonardo had access to Lorenzo de'Medici and the entire Medici family, including basically every single skilled glass maker, artist, scientist, and optometrist in Italy and was extremely curious so of course he would have had access to these devices. Point is, once something in art turns sour, that creates a stigma that lasts hundreds of years where even if you create that thing, galleries, museums, and critics will shun you. AI art was here from day 1. The real question is, what is generative AI attempting to achieve when the same thing can and has been done far better with photobashing and composites? The real AI tools for artists are things like Adobe Firefly. The rest of these generative garbage will be legislated into non-existence.
So what happens when a bunch of artists try to claim a style when multiple people share a style? With the art world being as toxic and dog eat dog as it has become this problem will inevitably arise.
So glad I found your videos! The algorithm introduced me to one, and now I'm just finishing up watching almost all of them in one binge session. Well done! 非常好看!
only solution i can think of is... the redbubble model... you have databases and artists uploading to them, ONE DATABASE... for each AI... so the artist gets commission... percentage of how much was used goes to each artist...
One of the basic laws of economics is supply and demand. If the average person can create an image by just typing in a prompt, why will any company pay a trained artist to create images? Why would the general public think image making is a 'special' & 'skilled' activity, like they used to think in the past. The craft of image making could be relegated to a "bit of fun for youngsters" and not a viable career path for those who genuinely care about art and its rich history. A major part of the problem is perception by the lay person, as they could assume all images from now on are generated by 'just a prompt', even if they are not. I speak as a traditional artist who has seen the advent of digital and 3-D art and has had to retrain & embrace the changes to keep making a living over the years. Now this?
loved loved loved ur video, and including the quotes from both parties, and I'm happy that u spoke of something that I do not see discussed enough... it's up for AI users to be aware of what this tool actually is, and use it responsibly, be it choosing to not use it at all, or at least be aware that it is based on the labour of all those who came before us, and how best it would be to use this tool to honor and respect their work instead of bashing all this and claiming it our own..
As an artist I can say that being scarred of AI is like believing in conspiracy theories, and again putting oneself in the mode of a 'victim' not a 'creator'. To be able to see AI as a creator I think it's amazing in this new era of these exciting new technologies to embrace this new vibe and mix it with one's creative process. Generative AI gives much cool ways to add to the process of developing of an idea. Rather than scrolling through Pinterest and get ideas from somebody's else art. Also would like to highlight the part of comparing mass market vase to craft vase as art and not being a product... hmm there is a big difference between art vs craft, craft is product and has a particular purpose to use it in day to day life, when an art is something that has emotional or intelectual purpose to make us feel or think :)
I'd also add there is a phrase by someone "If they copy you it one more time confirms your genius". And it also comes from the creators mode, rather than from victim mode. It's like be above it all, keep creating and chasing your dreams...
I'm one of those evil "developers." I actually agree with you, and I contend that AI doesn't create art. It can't be original and it has only aesthetic appeal, no deeper meaning. That said, not all human-produced graphics have any deeper meaning or originality. Most art produced to use commercially doesn't rely on a genuine artistic process, either - it's made to satisfy a corporate checklist and earn an income. All that said, even for art produced for commercial purposes, shouldn't be outsourced to AI. A human can make more meaningful connections and design decisions.
well big problem of AI art was that there is no any protection fence for human artist and no limit line for AI educating method. recently, U.S. court was judged that the copyright law will not agree of the artwork that made by non-human. This is will protect the leagal issue in a while, but it could be not the perfact answer. The big corporation and company are still supporting the AI developer for make cheap and useful AI to decrease the workers and employers. This mean, we can't limite the AI developing process with the large economic support. That's why law needs to chase the techonology development speed or go more faster than it to protect the people who are in the week protection environment. The law still needs to develop more to protect the people perfactly.
The fact that AI learns differently than humans is already a case for considering the way a database is used to be different of the inspiration atitude an artist has on works he admires and copy, and I mean by this that the whole point of the AI is that it 'learns faster' part. AI doesn't have acess to reality, it works as a logical system upon certain premises or axioms - humans do. We have to make sense of our experience to make sense of the world around, including other works of art that come to them as a given object as any other. When someone asks you to explain what a cat is, you'll be using the experiences you directly had with actual cats to do so, not a 'database' of second to third hand information - AI doesn't operate at that level, it only concludes logically a given data, which is the reason why those new technologies such as Nightshade are so able of fooling them. It's a computational mechanism, not a 'epistemical' one. Also, AI imitates a style not by it's true cause. Take for example an oil painting with loose strokes - an AI is able to imitate the manner without the access to a real brush or ink, and make it seems as if it were painted, when in fact it wasn't, is just a pixel analysis that resambles a picture taken from the original. If you know something about painting and already asked for an AI to make one in a particular style, you'll notice how strange normally it seems - they're normally too smooth, or its brush strokes are irrational. This is also true when it imitates a given trace, which is sourced on the mechanical predicates of our shoulders and wrists articulations, when it doesn't have any hand or arm whatsoever. It cannot considered as a mere 'copy and paste' copy, nor plagiarism of our usual kind, is in a level of mathematical imitation in various manners that human beings aren't capable of directly doing (if you take digital artists for example, which works involves pixels - though photographers might be included as well - it is as if the machine were able to copy their DNA/celular information there and transpose it, much more different than making an statue with clay). In my understanding is a copying machine by design and can't possibly stand in its own feet as it only produces second to third hand products (photographers and included in this), and in order to this be different (in any other regard as well) it needs much more than today computational capacities, it needs a way bridging the gap between its logical capacities and outside world.
Pues yo soy artista y creo que esa tecnologia es imparable, pero creo que nosotros los artistas tenemos que utilizar la IA como una extensión de la propia creatividad, empleándola para explorar nuevas ideas, o incluso como parte integral de sus obras finales. En este sentido, la IA puede ser vista como una herramienta que amplía las capacidades artísticas humanas, permitiendo la creación de cosas que antes eran impensables. al final el que hace la ai sigue siendo el creador, el que propone la idea, el que imagina las palabras para generar la imagen, eso cuenta como una expresion de si mismo
We give value to what we create.....time spent should not be considered but the message of the piece....what the artist intention of the piece.....most artist that is afraid is creating art....in a commercial way and sell art in a commercial.....honestly saying when i create something...its a part me that i dont others to have...but i have to set it Free....the value of it doenst lost of how much they pay for my work....but on how they see my value....THE art only devalued if ARTIST CREATOR chooses it to be so but not to the critics... art collectors....art lovers.....THE ONE YOU CAN HAVE MORE VALUE...IS A ART PIECE THAT AN ARTIST WILLINGLY AND SOULFULY GIVEN with no price......not in auction houses......Pricelessly given as a tokken of gift.....THATS all...i used A.i art.....
Well put, another thing I thought would be a good addition was the pop up of financial accounts across all platforms promoting AI to make money, if you just search AI art, you see enough small financial channels grifting AI art to make money. This not only dampens the motivations of artists and illustrators such as myself to promote against them. One person can run multiple accounts to cast a wide net while keeping under the veil of anonymity but also benefit from the present state of AI art being able to produce art that can appeal to the general audience compared to one person having to manage a single account on multiple platform. What AI art has also brought is this mindset of "artists being replaceable" now and that's a dangerous mindset to have, because of the grifters and promoters of these AI softwares repeats the same arguments, thus resulting in this oversimplification that people may not need to commission an artist anymore nor a company may not need to invest in 20+ artists/illustrators. This is a pervasive thought that has been shown true and continue to echo in similar form throughout these AI art debates in addition to the many you've already pointed which misses the entire arguments artists brings up to the people. Not only does this dehumanizes artists even more than we already have been, it is an attack on art integrity. People often forget that the past two years, artists were already devalued and exploited by NFT grifters. Now we deal with AI grifters and it's worse because the backings from these are funded by many large corporations to develop these technology.
You got great points but your argument was focus on visual art which leaves out of the picture other branches of human creativity which is why other artists like music artists, writers etc should pay more attention to this issue also. I think we're all artists in a way or at least potential artists so maybe everyone should at least be aware of the issue.
Paying for the "process" is like paying 10x cost for a cookie because it was made with love. That just sounds like bad judgement to me. Fine art collectors pay exorbitant prices for art because its a concocted tax dodge. They they come up with post hoc bs reasons why its so precious and how art is so meaningful. No, they are just hustling the taxman. But I think people who don't know better hear that stuff and take it to heart.
I have to disagree with this notion that keeps getting passed around about AI being a "tool". AI/ML Image Generators being cited as a "Tool for art" is extremely strange. A tool is used to aid in the process of a product. An AI/ML Generator literally just spits out an image. That would be like a chef going, "Hey look at my new cooking tool." and pulling into the MCDs drive through.
@@birdedex Wouldn't that assume that marketing of commercial products has any impact on how the tech itself can be used? I fail to see the relevance in terms of the argument that it can be utilized as a tool.
Great video! The solution was already established by the same companies that created these models. It's an issue of willingness not capability. The music models already easily avoid training on music protected by record companies.
@Cynthia Zhou I disagree with your proposal because, first, I think "in the style of" should constitute copyright infringement. That artist's complaints are misguided because copyright infringement must occur at the time of sale, not at the time of production, that is to say that if I "sell art" than I as the "seller" should be legally responsible for selling a work which is infringing. Second, just like with the music industry, the act of dealing with an artist is both expensive and time consuming. You are asking people to trade 5 mins and $0 for returning to "pay me $1500 per image for commercial rights and I retain creative control," when in reality even if I like the "stroke style" of some artist maybe I don't actually like "THEIR ART WORK," only some specific brush stroke. In other words, it's a pointless loss of control and extra cost. I appreciate your argument that artists didn't get their piece and that there is nothing to learn from if no one makes anything new, that's completely right. I definitely don't think copyright actually applies to perception, in other words machine learning isn't infringing copyright because it is simply reading the data, that's actual nonsense. However, the truth of the matter is the company or companies took a bunch of art data, which they definitely required, and then sold the result. Talking about copyright actually derails the argument, because it's ignorant and only makes the arguer sound like an idiot to anyone who understands the technology, without exception. Your magic paint brush example was spot on, it hits the heart of the argument and the emotion of the argument while retaining reason and the appeal of the argument is the infringement of capitalist liberty, that of the artist in their contribution to the creation of that capital. I'm convinced enough by it that my stance is almost swayed, it would literally be swayed if I wasn't afraid to give up my civil liberties, i.e. the right to view, hear, or perceive copyrighted works in private as a human being, the right to own my own 'data' or in other words, the right to own my computer even if windows or osx is installed on it, and so on. If you let the legislators decide on too many of such things, I fear where we'll actually end up.
There is a way to solve this problem, if there was perceived value in collecting vast swathes of images at a time to train an AI model… then compensate the artists
Remember when NFT's where first starting out, and the first controversy was that they were stealing art, monetizing it, and giving no credit or compensation? I think that's what's going to happen on mass. Especially by less scrupulous actors, and people from countries in which it's difficult to litigate
There's value in taking vast amounts of photos of public spaces and streets. That doesn't mean car manufacturers, architects and storefront owners are owed money by google for their creations/designs being in street view.
I really love this video and co-sign all of the points you brought up here. As an artist seeing this debate flare up, I came to the same conclusion as you did, just using different words. I've spent my time creating "art," as in, what you all would expect, a piece of me that I created that was inspired and expressed. I've also spent my time creating "designs," that is, purely graphical things, usually through software, that people and organizations need. I don't see the latter as "art," but I get frustrated when "AI Art" is being touted as "art" when it's just "graphics," or as you described -- process. Neither is better than the other, this is just arguing semantics. What I don't like is all the fear mongering and finger pointing that's happening, especially coming from artists. I get embarrassed with how woefully uninformed they are about technology and are just saying it needs to be banned flat-out. That's not going to happen, technology will keep progressing. We need to determine how we perceive this. In the perfect world, as you described, we could train AI to our own works and produce things we simply don't have time for, just like outsourcing in-betweens from a Korean animation studio. If money was never exchanged, this would be fantastic. AI being used as a tool. But the issue is money is involved, artists are being displaced, and those that own the software charge per image. If this was really a tool, as it should be, then there would be no charge. No one charges me each time I hammer in a nail. Unfortunately, we don't live in that perfect world. That said, plagiarism and stealing have ALWAYS been a problem in the art world -- we're just now seeing the tech bros give people license to do it through another degree of separation. But as you said, these bad actors would have done the same anyway. If NFTs, the metaverse, and crypto are anything to learn from, I would say that these grifters will be done in about a year and move onto the next big tech "disrupter." The general public will have a keener eye on how to detect AI art and will just brush it off as trivial digital bloat. Websites like DeviantArt, Pixiv, and other communities will establish guidelines and expectations when it comes to AI generated anything. Once all these scenarios eventually happen, artists will have a chance to take ownership of this tool and use it as it was intended. You can both hate the grifters and thieves while still advocating for the advancement of advanced computer generated graphics.
To Mrs. Zhou, and to all people in the comment section, I understand the concerns that AI brings about. At best, I am optimistic with the progression made on AI, but at the same time, I am very cautious about it at the same time. Given its advancement in recent years like OpenAI's ChatGPT, I am sure it is all impressive stuff, but what I have long spoken about, and what I do fully perpetuate, is that having AI replacing every single job in existence, is very reckless, and tends to cause more harm than good. And God forbid we ever tie an AI to a missile silo. That will be an absolute travesty, and it will be a massive indictment to self-destruction. That's why there should be red lines that should not be crossed when it comes to using artificial intelligence in general. At best, only most areas could be replaced with AI, completely or partially, but with the likes of cooking, policy enforcement (looking directly at you Google, DeviantArt, and TH-cam), and certain other fields, are better off not being used for artificial intelligence. At best, with the likes of editing and creating source code, they can be an exception, but by all means, they can make creating source code more efficient, and the editor just needs to edit other areas of the code so they can be satisfactory in the given outcome. In addition to the topic of policy enforcement like bots to content ID systems, which is what the likes of what TH-cam, Google, and DeviantArt had been using for years, is just as reckless. Because for a start, artificial intelligence in general, lacks a conscience, thus they cannot accurately make the best of moral decisions, especially when it comes to punishing people and taking down contents. That's why policy enforcement, and taking down content that violates the policies and guidelines, are better off being handed to people rather than machines. As for AI Art Generation on the other hand, I also do firmly believe we need more restrictive measures on those systems, like making them far more expensive to use, while fairly giving the person the right to own the art themselves. In conclusion, those things I propose could much better compromises, rather than totally disallowing them. I was talking to the ChatGPT AI about the general rising hatred and fear of artificial intelligences, and even itself, does agree with that I am saying. I will provide the answers it gave within a text file, with the additional steps to ensure the safety of society. Are you more than willing to do whatever it takes to take this all on board to be examined by government leaders within democratic systems? Thanks very much for taking the time to read.
You presented all artists as being against AI art and seeing art as a process rather than a product. As a professional artist and designer myself who works with many other artists in the industry, from my experience this viewpoint is a result of a sampling bias. It might seem that way at first glance, but in reality this only applies to the vocal minority. The silent majority who are happy to simply add these new AI tools to their toolkit aren't going to be nearly as vocal online about this as are those who are panicking that AI is going to take their job. I loved your metaphor with cheap nice looking vases at the dollar store. This is a good way of looking at it, as it makes it clear that we should all collectively prefer to live in a world that features nice looking vases at the dollar store, than in the world that exclusively features expensive handcrafted vases. In my opinion, having artists consent to their art being seen by an AI is no different from having them consent to their art being seen by anyone else. They consented for their art to be seen in general, the moment they published it. I know that AI has seen my art, and I'm quite happy to have contributed an infinitesimal amount to AI understanding art concepts in general, as I am to have any human be inspired by or learn by looking at my art. Many artists don't understand how AI works, and are under the impression that AI is taking their art and remixing it with other art to create new art, while in reality AI is simply looking at it and learning concepts. When I create art, I use various processes and tools and mediums, rather than being married to one particular process, medium or style. This means that I don't agree with the sentiment SamDoesArt expressed in this video of "combining all our life experiences with how we approach every single brushstroke". That's a politically charged statement, and I'm quite sure he isn't literally doing that, as then each of his brushstrokes would take days. Rather brush stroking is like breathing or typing, more of a subconscious process. And puring and combining our life experiences instead applies to our intent, to where we guide the artwork towards as we work, to what we envision or discover that we want the artwork to become in a given context. AI does not replace any of this, and in fact beautifully integrates into helping us explore the space of possibilities orders of magnitude faster.
If the value of human-created art comes from the process, then human artists have nothing to fear because people will continue to purchase human art for that reason. The problem is, people (generally) won't continue to purchase human art for that reason, because they put more value results. Now, let's say that "real art" is always process driven, and AI art is fake art. Great!! Where does this leave us? People using fake art instead of real art. 🤷♀ It really doesn't matter what we do with semantics, people will gravitate toward a perceived value.
Embrace this new technology instead of complaining. The future doesn't stop. You've been using AI features in Photoshop like CTRL + T or the clone stamp for years, and now you're invoking the noble meaning of art. Let's have less hypocrisy.
In my opinion your solution for this ai ethics problem is forgetting one thing: these learning machines are made for profit, they are not made for the artist, the are made for mass consumption (that's why their prices are dirt cheap). Just because one person in the ai field said that "ai is not replacing artists and that it is a mere tool", does not make it true. Huge corporations are the ones investing billions of dollars, they don't need artists since they got their artworks for free. So this is a very profitable, very eye-catching business model. You invest in the tech, you invest in a non-profit that gets you 5 billion images as a database then you use prompters as a feedback tool for the machine (free work and sometimes they pay to work basically). It's true that we as humans are to blame for unethical or bad use of the tech, but that does not excuse the tech bros that promote "ai is democratizing art" speech while destroying the market and art communities. And I swear the "ai is here to stay" line is getting reaaaaaaly old. It only discourages people for fighting for fair use and steers towards complacency. Ai as it is now is not ok, and yes, it's the opinion of the few atm. But tides change.
It's true that in this capitalistic society, AI art is very profitable and that's why corporations will jump on it. That's why people like us are here to raise issues about it.
Also, I disagree that "AI is here to stay" discourages people fighting for fair use. I do agree that some people have adopted a nihilistic approach, ex. "AI is here to stay so what's the point." However, "the point" isn't about the end result, it's about the process of how we get there. And it's up to all of us to educate other people to understand it's not about the end result. It's about how we get there and how we set up a good foundation to support everyone in the process.
So quick explanation here, I am an computer science engineer, in university I have been studying deep learning and testing some hypothesis, i killed my old laptop by doing so, ai technology is very complicated when it comes to hypothesis with almost about anything you do with it. I did some testing with vqgan and diffusion models when they did come out to answer a personal simple question. Can ai be considered creative. The short answer is no, it is not creative, but please note that it is my testing i might be wrong with approach or execution. So i think you already familiar with diffusion models and how they trained so i won't explain that and straight explain what i did. I trained my model from scratch it didn't know anything, i fed him pictures of very stickmen alike black dogs in a white background with resolution 64x64 and amount of thousands images, i know resolution seems low but it is actually enough to test, and also fed him images of dog's head. And when i requested to cut images of head from dog images it did fairly good job not perfect but it was going in a right direction. And then i fed him images of blue swords and also images of dogs holding a blue swords with a mouth and images of red pistols without dogs. Now what we have model that can recognize dog, dog's head, blue sword, dog holding a blue sword, red pistol. Now moment of truth if i will request him do generate image of dog holding a red pistol with a mouth will it make it or not. And as you might know answer it just give up and were generating mess trying to blend together dog and pistol. So why am i writing all of this, i often hear arguments against and for ai art generation. Some people say that ai is training on data same as people do so it is not stealing, this argument is hard to disprove because nature of ai makes hard to prove or disprove almost anything but we can say for sure that humans are creative and as i tested ai is not, so we already process information in different ways, ai's pattern recognition works in different way than humans so calling out ai for stealing might be true but then again it is not coping it directly, it is using it is knowledge of pattern that just differs from humans. Okey i will finish my observations here and further will just speak my mind about this situation. Ai is perfect representation of our society. Have any of you noticed that every time something original happens like very successful game, movie or some song. Industries and companies will do everything it takes to take piece of this successful cake and they will deny any other ideas whether they are original or not. Most big companies in entertainment industry always make same thing over and over again and the only original works were made are always made by small group of people and when they reach their success Industry giants will start to manufacture idea of this small group in a greater scales to greater profit because it proven to work to profit. This ai situation is exactly the same, if you lost your ability to distinguish ai made art from human made it is not because ai is making original content it is because people are stagnating and making same stuff over and over. Ever wondered why ai so good at generating anime waifus, it is because they are very similar in appearance to each other because we stopped pushing it forward. By making stable diffusion open source we effectively made every human being a corporation, now every one can mass manufacture popular art or art style but nobody will make original art, and even if somebody will make original works it can easily be taken and also mass produced without consent or mention. Ai is here to stay not because it is better in art than humans but because it is better in worst thing about our society today at creating same idea with low effort.
This is very disrepectful and very ignorant to open source ai developers to like Stable Diffusion . Entire concept of ai art actualy comes from understanding how brain detects some specific object in a image.
@@lazarus8453 Ai in the beginning mean to mimic human but every day ai you see now is not, ai word it self that is used today to describe models and other tech is a buzzword that companies use as a marketing tool. All the ai tools that used now do not mimic nor similar to human brain, most of them is variations of interpolation and other math formulas, there is a several limitations that ai can't physically do or achieve and probably never will do to being complex math rather than brain. Of course if they change the core principles of technology itself which I don't thing will happen in near future.
@@cynzyyy100%
The thing about ai art is that it cant actually "add its own touch" because its not sentient. It finds trends in composition, colours and details and replicates those things based on the works under the keywords imputted by the user. It doesnt have an understanding of why those details are there, it doesnt understand 'people have five fingers' it understands that there are several of these shapes in these places.
Ai is a tool made for companies not for artists, and while it can be used to help speed up the thumbnailing process or nail down a composition or colour palette, unless youre an artist that only does one kind of artwork thats all it can do
I've been a freelance graphic designer/illustrator for much of my adult life. I've done a lot of commercial work for smaller companies, and I'm certain that most of that work will be done by AI in the near future. I think it's simply going to be substantially cheaper and faster for a startup or grassroots organization to get professional-looking artwork that way.
The issue I keep coming back to is that these companies will get EXACTLY what they want from an AI, no questions asked... and that's not necessarily going to be a good thing for their businesses.
I'd argue that the best creative work of this type comes from the push and pull that happens between a business-oriented client, who may not have a very refined artistic or branding sensibility, and an experienced artist/designer who can help make the company's unique vision & identity visually cohesive.
Even now, stock photos, generic vector art and design templates are already "good enough" for a lot of small to mid-size companies, so maybe AI just accelerates the trend of homogenized design for all but the biggest companies. Or maybe we'll see some genuinely weird/tacky branding based on the whims of a startup founder after a boozy weekend with Midjourney (that would at least be fun)? Either way, probably much less commercial work for all but the most prominent artists.
I totally agree with you, both with the direction this is going, as well as AI-generated "art" not truly serving its purpose - yet on the surface it seems to do the trick. But it doesn't.
I always think of the "art" in an image (or music, etc.) as the way it makes the audience feel. It's all about conveying this particular feeling that you want to convey. AI "art" cannot do that, so it uses randomness instead. Because it doesn't understand "feeling". It intrinsically can't. It was merely trained on images with a description of *what is in* these images, not how the image itself *feels*.
This puts AI generated images into this uncanny valley: it looks alright, acceptable, but something is wrong with it.... and we can't quite put the finger on what exactly it is that is wrong...
... yet the executive, who only cares about minimizing expenses, will say: "yep, good enough!"
So keep doing what you are doing! They will only get EXACTLY the effect that they want if they go through that creative PUSH AND PULL! :)
its the same thing that happened with video games, there will be little to no innovation by a bunch of rich kids wanting to make a quick billion, which will happen because people are deep in depression and alienation thanks to capitalism, companies are already grey matter and look the same to all of us, less people will question the continued abusive practices and the true citizens (upper class) will not give a damn and push their bland products onto all of us, normalizing AI made "stuff" fixed and catered with an ever diminishing skeleton crew, always happened, always will
In my 10+ years as an artist(3D Animator), all I've realized is that new technologies don't really make our lives easier. As soon as we are able to make things faster, companies just ask us to do more to make up for the time saved.
It would be great if AI could do the tedious parts I hate doing, but as things have progressed it's become clear that won't happen. Basically nobody in the AI/ML community has any clue about the intricacies of my work, and doesn't care to. It's easier and cheaper for them to just automate my job entirely and all that will be left is the tedium. Thankfully AI still isn't anywhere close to being powerful enough to meaningfully threaten my work yet, but I hope by that point I hope to be financially secure enough that it won't matter.
Also, one thing thing I feel should be pointed out(that other commenters haven't) is that AI/ML can only create something because humans did it first. Meaning that AI/ML "artwork" is inherently derivative in it's current function, therefore limiting the creative variability and leading everything to slowly become more and more homogenized. A problem that already exists, but I feel like AI/ML will make 1000x worse.
Something I haven't heard anyone talk about is saturation. I would think that there will be a huge saturation of art in the next year and we wont know if it is human or ai. Or if a mediocre artists start making hundreds of new art works. Multiply that by millions of artists. Does having billions of new original art pieces devalue art in general?
Great point! I do believe there will be a mass devaluation of art because of AI turning this into a "wholesale" process, but I don't think that will change the "value" of art, it just means art will become accessible and have higher production quality overall
I have also thought about this. But also what about Substantial Similarity copyright. When millions of art or graphic designs are generated every month. And traditional artists would constantly have their art removed because they can't make pictures in hours or minutes and therefore constantly run into copyright claims from AI artists.
I think a further saturation within the art space, where there is a much heavier emphasis on “art as product” and on “art for sale” will fundamentally change the mindset of artists, thereby forcing them to think more like salespeople than like artists that are making art primarily for the sake of making art or therapy and are not leading PRIMARILY from a profit making mindset. This would be a major shift in mindset and in the soul of art.
It could unfortunately start resembling the NFT space from back in 2021 and could fundamentally change art for the majority of artists outside of the celebrity art auction houses and for profit markets. Let’s hope not.
Saturation of any product devalues the product no matter what the product is used for. NFTs were supposed to give artists ownership of their digital art. There was already so much theft of digital art even before Ai. I had my art directly stolen. So I helped make a collection that I was very proud of with the hope that it help us digital artists.The scammers moved in and destroyed the whole thing. I can see this already happening to Ai imagery.
An artist should have the opportunity to make a living with their art. Unfortunately that does not look likely anymore and the more you work to create a distinct and recognizable style to more likely you are to have scammers steal it. I am sixty years old and I have seen art evolve in many ways. I have adapted and embraced all technology that has come along. Ai is different. It has given scammers the ultimate tool to steal and dilute the art market. This is the first technology I can't get behind. I already see it tearing apart the art world
There is already saturation in art. Everyone's and their dog are posting paintings on instagram. It's impossible anymore to stand out as an illustrator. AI makes it worse, but the problem is already there.
Okay, as an artist, there is a really big misunderstanding that's going around about how artists begin their process to become an artist. I hear a lot of people say that artists simply 'copy' other people's art in order to improve. This may be the case for young artists, aka children, but for a lot of artists in the community we stick to a lot of rules in order to be respectful of others' art. For example, we may copy elements of an artists processes like colour, composition, or linework. These things are simply fundamentals of everyones art that everyone incorporates when making a piece of art. I hear non artists say a lot that we simply copy the whole art in order to learn but this is something that is very frowned apon in the art community and never acceptable. As artists we believe an individuals style is created by external elements like upbringing, culture and also other artists processes. All of these elements are what make another persons art unabled to be copied because it's unique in the way their brain and hand processef that information. In my opinion that's what makes art, art. Please be more careful when talking about the comparison or how an artist vs Ai learn becayse it's not even remotely similar and that's why AI art always looks like a copy of someone's original work. When i look at AI art i can tell directly the artists that were plagiarised to make the 'art' and just because you can't doesn't mean it's original work made from 'learning'.
It doesn't take into account the failed artists.
Yes, all artists copy and take influence from other artists as they are forming their OWN style and voice. The artists who fail to create their own style who do not have their own unique voice coming through their work, are not successful. Their work is called a cheap copy, a knockoff. Not everyone who aspires to become an artist becomes one, in fact only a small fraction does. This is where the struggling artist and failed artist tropes come from.
Just because someone can copy the work and style of others, it does not mean they are a "good" artist. AI art is similar to the artwork produced by the people who do not have their own voice, who can only copy the work of others, and not be able to convey their own emotions through their work. It is hollow.
@KNYD Yeah, I completely agree that AI art definitely has the look of a failed copy, but in my personal experience, I fully believe artists do not and should not copy for practise. Simply referring to copying the entirety of the piece from the lighting, composition, and linework just everything. I don't like people referring to artists practising by imitating an ASPECT of an artwork as simply 'copying'.
Also I think the failed artist trope doesn't really come from people who can't make their own unique look but rather artists that have a unique look but struggle to be noticed or simply from lack of pay that is still a very prevalent issue. But that's just my opinion, it could be different for others, I suppose.
@@KNYD Well, that's major BULLSHIT and/or a lie. Not ALL artists who achieve success are truly original or have their own thing. And not ALL the artists who fail, or simply don0t become mainstream is because they don't find their own style, I WISH It was like you say, but that's BS.
If you were right, the majority of people would know who the fuck Neil Gaiman is, and not so much who JK Rowling is, but that's not the case. And I use SPECIFICALLY those two artists, since is well known by real geeks, not modern millennial ones, that "Harry Potter" is basically a COPY PASTE of a 4 number comic story by Neil Gaiman called "The Books of Magic"... But, you didn't know It, did you??
It’s simple the ai cannot think. It’s not thinking or making active choices like artists. To say a machine is the same as human is stupid.
We are conscious. We know and are self aware. Look at hands. We understand it’s a hand. We know what it is and how it works. The ai only sees patterns and symbols. It will need images and 3d models to know what a hand is but we know what a hand actually is.
Saying that digital illustrators are "not really creating creating" is pretty uninformed and unfair. Digital painting and illustrating is actually not that different from traditional painting and illustrating, you just have different options and limitations, but the creative process is in many ways the same. And saying that AI is just a tool just like photoshop or digital camera is also very much not true. This is the first "tool" that could potentially replace the artist entirely in majority of cases, thats why it really has a potential to destroy a lot of creativity and a lot of dreams.
The tech and all models must be strictly open source to the public, and companies like Stability AI and Midjourney shouldn’t have the right to profit from the sensitive data of others.
When people say AI art is stolen art, it isn’t quite literal but there is a truth to it. It is in fact stolen labour, and there are companies that are using the stolen labour of hardworking artists for their own profits. That is indeed theft.
No.
People need to realize that "quicker" doesn't automatically mean "better". I'm so sick of everyone saying the "old ways" are "obsolete". That's what they said about vinyl records, and physical books, and they're doing better than ever. I don't think the other tools are EVER going to go away. We all have the choice to _not_ use this technology if we don't want to. Tools should be seen as "options", and nothing more.
When it comes to this stuff, I don't have that same level of appreciation as with actual art. It's the equivalent of a robot climbing Mount Everest vs a human doing it. It's cool that the robot is scaling it, and accomplishing so much in a quick amount of time, but that's what it's designed to do. When a human does it, it's much more impressive because we're flawed, fragile individuals, and it's so much more satisfying when they actually reach the top.
Overall, I'd rather wait for perfection than haphazardly slap on an image and call it a day. The thing about "instant gratification" is that it's there one day, but then it fades almost immediately.
Tbh its just wrong from AI fanatics to say there are no unique art styles anymore and everybody is copying someone, i would consider my artstyle to be very authentic and different because i am only following what my mind comes up with. I am more process driven because it speaks the inner language.
Straight up there’s art styles that are unnatural for some people to mimic because of their pre developed muscles. I did twirling for years and my art is naturally very curvy trying to say draw an original character to fit in in a show that is very triangular isn’t going to work it just wouldn’t it’s going to have a noticeable fingerprint
TH-cam recommended me one of your videos and now I'm binging all of them. I can see how much time and effort you put into each one! New sub ✋️
This went a different direction than I thought. It’s a lot better than many of the other videos, but why ignore legal action or etc. why not show that without the massive amount of data it couldn’t depict images to the same quality. Why did you ignore the huge amounts of differences from human learning and machine learning even as a lampshade would be something. It sounds and uses the language of supporting artists but is strangely in deeply support of AI, basically saying throw our hands up and do nothing but call out bad individuals. Instead of regulation of a dangerous “tool”
Also legit question did you ask/talk to the artists you included in this video?
I’m inclined to agree with those who have pointed out that your solution would require a large shift away from profit motives in order to work. The problem here is in reality not much different from those we face all over the economy. The issue would be of almost no consequence if it weren’t for the income loss to artists. With a corporation creating these art factories and keeping all the money, we end up with increasing wealth inequality, which is possibly the most consequential aspect of our current onmi crisis. Every artist whose income dries up is a cultural loss to all of society. It may not be one you notice, but add them all together and we get a shrinking pool of human creatives left to bring the truly novel ideas to light. Champions of the tech are too quick to embrace the cynical idea that these machines are the same as people. And like every sector that derives profit from shedding workers, they forget that if everyone is broke they can’t buy their products. If any of these companies want to bring about fully automated luxury communism, be my guest. But from what I can tell they’re quite glad to exacerbate and exploit a growing servant class. And even if that weren’t reprehensible, it isn’t sustainable.
There's a lot of misunderstanding of art in this video,I feel like the biggest one is that an artist doesn't just copy other artists to learn. The difference between artists and AI is that artists can get inspired by anything, not just other art. Many artists don't learn by consciously copying anyone's work. They take things from here and there to build their own thing. It can be a cartoon a photograph, a really cool shaped tree they found on a hike, it can be anything. Comparing the 2 processes like it's 1 to 1 is insane to me. They are nothing alike.
Having artists use AI solves a non issue funnily enough, AI art itself solves a non issue, artists make art because they want to first and because they profit second. The art industry is awful to 80% of artists out there, most people cannot make a monthly salary doing art, yet they do it anyway, because they want to.
As someone passionnate about both art but also how quickly technology is growing, this video was the perfect length 😌
What I really mind is that art creation is a big part of my growth.
I started drawing since I was like 8 years old and at this point is something that I really love, I want to make something I love, something I enjoy.
I just don't want to have to stop making my own drawings because "AI makes it better and cheaper"
Every child draws at some point, or colours or otherwise wants to create, it's something I think is hardwired into us, some people grow up and move on, others find different ways to express themselves, and some stick too it for life.
I like drawing as a hobby, and I don't plan on stopping because AI is faster or cheaper, and I don't want to give up on human artists ether, if your good at something you like doing, and can get paid for it, that is a win all over, I don't see less money going to individuals, and more to the AI service providers as a ultimately good thing.
Who says you will have to stop? If you love it there's nothing stopping you from doing it as long as you're able
Just because ai art can make art doesn't mean I wouldn't pay an artist to create something that I want. I can also get a more exact piece of album art or whatever from the artist. AI is hit or miss. It's cool but it takes a lot of time and prompts to get what you may want.
I am very late, but the AI art issue is about two things. One is the understanding of what is theft and if AI art is stealing, and the second is artists being threatened by a more efficient method of producing artwork. Something lots of people don't understand is that AI art doesn't use stored image from training data, it uses parameters based on noise maps made from images, which look like static. The training data is terabytes of size, while the AI itself is only gigabytes in size. Their artwork is not being sold for profit, as the images made are so far removed that they can't be considered identical, and are all transformative. This is why copyright lawsuits against AI art generators have been failing consistently. Artists are threatened by AI art because it can make products of immensely high quality for practically free. Artists will maintain a niche, and some have even benefited by raising prices as a response to AI art, but now art creation as a whole is accessible to the masses, rather than locked behind a paywall that feeds into artists' incomes, or behind a massive time investment to become skilled at it. Also, the point of "unfair capitalization" seems odd, as there seems to be nothing unfair about developing a better product to accomplish a task. Otherwise it would be "unfair capitalization" to use excavators, bulldozers, and cranes in construction, rather than 100% human labor. Also, you can't claim a style, it's simply not possible legally, an arbitrary line anyways. By claiming styles you would bar basically anyone from drawing in various styles, preventing people from making comic style and manga style art as an independent creator, or turning "How To Draw" art books into legal traps. It is an unironically tyrannical concept which would be an absolute disaster for the art world.
Also, 19:33 that's what she said lmao.
I think the issue with being able to copyright art styles is that, it then means you will have a lot of problems in terms of other artists suing other artists because their style is too similar to their own. And how on earth would that be measured? How can you look at it and say if it's too similar or not? What would the determining factors be? This is likely the reason an art style can't be copyrighted as of now.
In my opinion, the way it would work best for both parties is if artists simply get paid for their work. Let's say an artist knows their art has been included in the data set (though I do believe it would be better to be opt in). They could be able to go to the AI tool, and make a claim and with validation that they are indeed the copyright holder, they would then get compensated for every generation that used their work. Basically, they would get a certain percentage of whatever income the tool makes.
You know similar to how Spotify works for music.
I mean, yeah, that would be better than how it is now, but right now there's already an issue with AI not being able to unlearn what it analized from an image, so it still gains something from it even if it is deleted.
The otger issue is that spotify SUCKS for artists, they pay the least out of all the music streaming apps with maybe better pay per stream for big/mainstream musitions, the thing is, it's kinda functioning as a monopoly.
I do wonder, however, what if there will be a new category of enforcement of creator's rights: artists could copy an artstyle BUT it can only be enforced against AI genarated images.
But the artists put them out for show on a publicly accessible site to begin with.
@@_loss_
Not entirely sure what you are trying to say. Putting something out to show doesn't mean it doesn't have copyright protection.
So for example, if you see an image, sure you can view it, and download it. Nothing wrong with that. But if you sell it, or use it in another project and sell that work... then depending on circumstances that can be considered copyright infringement.
The question really is, does using copyrighted works to train an AI for image generation count as fair use. We will not know till a court case occurs and the court decides.
An Ai can measure how much and what styles it uses and tell you, but it doesn't and AI users probably won't show what prompts they use if they are using someone else as reference.
How come an artist can show their timelapses and references and an AI user won't be as transparent?
@@anhsu4854
How does any of that have to do with what I said?
Or are you replying to someone else?
As an artist, I do believe art to be process driven, and yet, I don't believe that AI art lacks a creation process. AI democratizes the creation process of art and gives that creation process to people who, for whatever reason haven't or even may not be able to go through the steps of traditional training. Not everyone has the mental or physical capacity for that kind of training or time to work towards it.
Who are we to gatekeep methods of creating visual expression through text based generated AI images? How can we say, your imaginative creative expression of yourself facilitated by technology is not art?
The issue is that AI won’t displace artists careers, such as Concept artists if the databases that the AI models like Stable Diffusion are built with didn’t steal millions upon millions of images under the guise of research purposes that were then conveniently used commercially.
Yep! Similar AI models in the music industry are trained on copyright free works.. because otherwise they'd be sued. They have issues with overfitting meaning that sometimes you can find a whole chunk of someone's song in an AI song.. THE SAME issue that the AI art has. Look at half assed signatures in the corner that the AI tries to replicate, for example.
The law industry is a joke- supposedly to protect everyone's rights, but in reality, only if you have the money to jump from court to court as a multimillion company tries to bankrupt you to give up.
If you make a collage that you then sell, you can't use copyrighted images for that. Even for photobashing there's bundes of photographs you can buy to legally use. When you use copyrighted work to make a machine that you will sell licences for and the users will sell artworks from.. same applies, imo. Sure, the art is "not in the finished art". If I stole a truck to build my house it'd still be llegal even though the house doesn't contain the stolen truck.
Bro, you have a copyrighted character as your avatar. You can't lecture anyone about stealing IP.
@@josehumdinger6872 Bro, he's not making money off of it and if he did, he would be sued - can't say that about the AI database which is obviously also fed with IP-data if it can react to a prompt referring to an artist. You win dumbest comment of the day
@@KarimNiazi educate yourself how copyright works before you keep embarrassing yourself.
In the cover of your video, it reveals two version of your portraits. How much it cost?
AI companies used artists works included some of my work which was in Laion 5B to train a commercial use AI model. They didn’t reproduce or transform any of the files in it but directly feed those copyrighted artwork without consent.
Then, the machine “deep learning “ it. How do people define its “learning?” This word is only that scientists created to let people understand how it function easier and trying to say that! “Hey! Machines can mimic what human do!”
I’m an award-winning illustrator and spent years to develop my style, but AI companies scrape my art to train the AI without my consent.
I can actually create a unique portrait, but now, will people pay me instead? A digital quick portrait is minimum 200 USD for only personal use. Every generated image for profit use means one human artist with that style short the money to keep them surviving in this unethical world of AI.
Accordingly, it is not a tool for artists now, and AI companies are did it wrong from the beginning.
Those people are still build the lies to make people hug and accept this unethical tech…
Yeah, it's mostly nonsense to not see this as a blatant violation. It's just impossible to have the vast database without doing unethical stuff and furthermore as you correctly point out, the idiotic reductionist claim that a machine "learns" and the equivalency made between a human process and a machine process is mind-numbingly evil.
It's sad to hear about your situation. I switched to motion design because the illustration/art market seemed so crazy but I know those mega-corporations psychopaths are gunning for me too. I'm thinking of switching careers, let those fu** rot and changing to physical art (painting and sculpture) without the aim of commercialization. I just don't trust people in general, they all talk lovey dovey but if money is involved most won't value art and craftsmanship.
Excellent video! Best explanation of AI Art ethics problem. The magic paint brush was a hilarious prop. I thought the video was very informative and not too long.
glad you enjoyed it!!
Legally you can't copyright style so you can't license it out & have it be protected.
Copyrighting art styles would hurt artist to
Personally, as an indie gamedev and game design student, its gread for creating general moods or ideas and its especially Usefull for fast-paced prototyping. In the end if your making the final product you will still need artists. The artworks created this way never see the light of day. So i feel its great for getting inspiration, same as going online and looking at art pieces for inspiration, but i can controll more factors.This is a very difficult complex topic, and we should add to this discussion. Its not black and white, but AI art defenetly has its risks.
The cited Reddit comment at 4:13 shows the problem pretty well, IMO. "Anytime 'good enough' is the art requirement, people will be using AI".
Your videos are next level!
I had never thought of art as a process before today. I’m not typically a fan of longer videos, but this was well worth the watch.
Digital illustration still has a process behind it. It’s still art. It’s still creating. It’s not any less valuable because it’s not traditional art. Insinuating such is honestly really offensive.
The problem with art appeared after WW1, when dadaism claimed that any random event is art. Well computers can do more beautiful random events. But what is missing in computer generated art and in dadaism, is intent. An artist made an art to show us something, the generated art is something that is happening without intent, is like blooming of flowers and drift of clouds. Only when the flower is drawn is becoming an art object.
Honestly, you’re doing a great job with this channel. I admit I will try to copy some of your editing choices! Keep it up 😎
Thank you so much!!
@@cynzyyy Just trying to be truthful, Cynthia!
So while I agree that in general AI is here to stay, I do not agree that this particular use case, application, and business model is here to stay. When you speak of the fact that it's an impossible problem to solve to try to credit or compensate each copyright holder in a dataset, that is tuff titties for the developers, not the general public. "We'd love to NOT steal your work, but it's just too hard and our technology MUST move forward" No, I'm not buying that. Here's some, IMO totally reasonable ways this can move forward:
1: AI can use exclusively images in the creative commons, and artists who opt in. You can create a myriad of incentive models to get artists to opt in, that's a huge discussion, but that's the main idea. Almost certainly, this would result in the model not being as good. The model not being as good doesn't prevent the tech moving forward. It ONLY stops it from freely exploiting the property of people.
2: At any point in time, artist should be able to blacklist their names from being invoked in image generators. I can't think of ANY legit reason to not have this be the case other than companies specifically wanting to exploit and steal from artists. I am not a genius. Someone smart has thought of this before. It's NOT being done for a reason. It is MEANT TO STEAL.
A few additional points -
This fight is currently about art, but everyone should be in favor of some common sense protections about the usage of data for AI in general. Right now this might seem benign to the average person who all in all doesn't care that much that this is stealing from artists. But understand that everyone has images of themselves all over the internet as well all over social media. Lots of it conveniently tagged with who is in the photo. What is all of this was put into a model and trained to be able to invoke your face by name? If anyone could create images of you, your children, your family, doing literally anything. If anyone could take your face and make and sell merch with it on it, then tell you they don't owe you anything, that your face was freely available on the internet, that it would be too hard to NOT use your face.
Now consider that for many artists, their art and their style is more central to their identity than even their face. There are many famous artists out there that hardly anyone knows what they look like, but they could recognize their work a mile away. It's not just someone's "product" or even their property being stolen, it's their whole identity.
>If anyone could create images of you, your children, your family, doing literally anything. If anyone could take your face and make and sell merch with it on it, then tell you they don't owe you anything, that your face was freely available on the internet, that it would be too hard to NOT use your face. Now consider that for many artists, their art and their style is more central to their identity than even their face
You are mixing up one's right to their image, their brand, which is a long established right - you can't just create a fake 3D Tom Hanks and put into your movie and say it is Tom Hanks.
You are mixing that with the right of one owning an art style, which do not exist: anyone can draw like you, art style aren't copyrightable. They never were (Disney would love that art styles were copyrightable btw). So even in a world where artists "got their way", and you couldn't use other people image without "their consent" to train a database, there would be nothing preventing companies from just hiring some skilled artists to copy your style and than pay a few bucks a single time for having the right to THEIR paintings, and use them to train the model.
Also, this world where "anyone can create any image" pretty much already exist. Anyone can go to your facebook, or instagram, and download 10 photos and use Stable Diffusion and Dream Booth, fake any image they want. This world already existed even before that, with the invention of Photoshop, it is just easier now. It is all open-source and freely available, the genie is out of the bottle.
@@someone1861 I'm not mixing these things up. I'm simply making an analogy.
In any regard the artist doesn't need to own a style. They own THEIR WORK. That's what's being used. It's not "their style" that's being used, it's their actual work. The generators interpretation of their style is the output, that's a product created. A product created by directly using something that the company doesn't own. Each person has the right to not have something they own be used to create someone else's for profit product, or at least they should.
@@midlet Copyright isn't actual ownership though, it's a limited monopoly over reproduction rights and substantially-similar works. "IP" are not equivalent to physical property (and there's even a strong case to be made that the legal construct called "IP" infringes on physical property rights!)
I could also see some form of dystopic logic being applied here in this thought experiment: If you listen to music while you make art for a client, the music is influencing your mood, your thinking, and thus your art. Does that constitute "using something you don't own" to create something else for profit? I would hope that most people would see this as a ridiculous standard to hold, but there is probably some executive in a record company who dreams of such scenarios.
Your "AI Art Proposal" would unfortunately have serious negative consequences, despite your best intentions. The main purpose of copyright law is to promote the creation of new, original works. By restricting the use of specific styles or techniques, copyright law would inhibit creativity and innovation rather than promoting it. Many artists learn and grow by emulating the styles of others, and some styles become whole movements that define an era.
Additionally, art styles are often abstract, fluid, and overlapping. They evolve over time and can be influenced by various cultural, historical, and individual factors. Defining an art style concretely enough to be subject to copyright would be extremely challenging, if not impossible.
And finally, copyrighting art styles would lead to the monopolization of certain artistic expressions, which would be against the public interest. It would limit the variety of artworks available to the public and potentially stifle new creations.
The problem is: why are so-called "Artists" an exception? AI feeds from everyone: from actual photos taken by photographers, to generate photorealistic art. From code written by programmers, to generate code. From articles written by writers, to generate text. From scripts written by screenwriters, to generate new scripts. And so on. This will hit everyone and NOBODY is exempted from this.
They’re not, you probably only hear from them the loudest because of what’s fed to you by the algorithm. Everyone you listed is on the exact same side and is having the same concerns and problems. It’s not artists begging to be exempted, it’s creatives and workers asking to not be exploited and profited off of without their consent or simply disposed of.
This doesn't take into account the failed artists.
Yes, all artists copy and take influence from other artists as they are forming their OWN style and voice. The artists who fail to create their own style who do not have their own unique voice coming through their work, are not successful. Their work is called a cheap copy, a knockoff. Not everyone who aspires to become an artist becomes one, in fact only a small fraction does. This is where the struggling artist and failed artist tropes come from.
Just because someone can copy the work and style of others, it does not mean they are a "good" artist. AI art is similar to the artwork produced by the people who do not have their own voice, who can only copy the work of others, and not be able to convey their own emotions through their work. It is hollow.
I don't think AI art is an apocalypse like so many people make it out to be, since people have show time and time again that people like Human made things.
You can make incredible music without hiring a single musician and have been able to for well over a decade, yet musicians still exist because people like human played music even if it is possible to recreate it entirely with computer programs.
Physical calendars are still selling well despite digital calendars being a thing for ages.
etc, etc.
People like human made physical things. Wait a decade or more and you'll find that AI art was not the apocalypse so many made it out to be, Human made art will always be popular.
as an artist, I feel like you were missing the point that most other artists have with ai. its not the money making element, bc that is inevitable as you put it. its that these ai are being trained by non public domain images in the first place. every image on the internet belongs to somebody, and the people who made them should have been asked if they wanted to opt in or out before the ai was trained. then you have a complete list of who's art is in it. if im not mistaken, sometimes ai will spit out an image that's a direct copy of someone's style, even if the person didn't put in a prompt asking for it. but only the artist or someone who knows their style well would be able to tell, and the odds of that happening are slim to none. this whole video I was waiting for consent to be brought into the conversation but it just never was
> *missing the point*
TBF there isn't some singular "point," many reasons, or combinations of reasons depending on who you ask.
> *by non public domain images in the first place.*
IMO that is kind of an incorrect pointing of the problem - since in a few countries, including the US, anything eligible for copyright is copyrighted, and that includes works you volunteer or allow, and creative commons works. It confuses licensing status (and whether licensing is needed) with copyright status as if they are synonyms, when they are not.
I think your take on the ethics of the AI I think miss a couple things. First, the artists who published their artwork online do have a license associated with it. If that art is then used to feed the training of the AI in violation of the license, then there is a very real legal problem with the people who chose the training data. For example, if art is just public domain, then there is no problem. However, creative commons style licenses often have attribution clauses and/or non-commercial use licenses. The generated AI art doesn't have the necessary attributions. The big challenge here is a legal one. What constitutes a forgery now has to be refined. It's not the first time that new technology has forced society to adapt. _Some_ of the same arguments were applied to photography. The advent of AI art, which is a very new capability which is going to force our society to adapt. It's an open conversation, and we as humans are still trying to come to grips with what should be the ethics of training AI/ML, and the implications of what that new capability provides.
all countries work differently, sure you have copyrights to your artwork in most countries, but also art can be created from other artworks as there are many such cases. It just has to be different enough.
the long vid format was nice, thx for explaining
I think the only way for this to work and it not screw artists over (because some of the options such as oh you should have something like copyright over your creativity could cause things like people having to pay photoshop when they use their brushes) is just the requirement of consent. Consent for every image the thing is trained on, for that consent not to extend to other ai and for no ai to be comprised of more than 15% of one artist’s work (or that if it is they are given partial credit but would have to agree to every image used) and I can hear people saying’but it’d take weeks or months to do that!’ Yeah but months is shorter than the years or decades it takes to refine your art from scribble to masterpieces
Artists can have whatever opinion they want about what defines art and what makes it valuable to them. But as soon as you bring in a consumer the artist's opinion on the subject isn't the only one that matters. Consumers get to choose what is valuable to them, and what they want to spend their time and money on. And this can't be an either or where AI art isn't allowed to exist simply because it brings competition and possible loss of income to traditional artists. If that's how society worked we wouldn't have cars because we have to protect the livelihoods of horse breeders and buggy makers. AI art gets to exist and traditional artist will have to adapt to having that as a fact. The only part of the artist's argument I can get behind is AI directly using IP of other artists to create without giving credit or compensation. I say all this having been an artist all my life.
Excellent breakdown and suggestion for how AI could be used ethically. I know it would be immensely profitable if a famous artist with a distinct style (Sakimichan or Artgerm for example) sold an AI engine solely trained on their entire vast body of work, and clients can pay to use it to generate quicker, smaller commission work. The AI would also automatically print a watermark or signature on the final piece after processing so people know who it was sourced from. And both sides benefit too - the artist owns their style and monetises it, and the client is able to purchase access to get something in a style they love while supporting the artist.
Yessss
Thats not making any sense..if a machine can produce sakimichan artwork legally in mere second, then the real sakimichan is over.
@@tokosjr7560 no bc they'd have to pay the artist to make the program generate their style
*Context of my comment:*
I am somewhat of a hobby artist, I like drawing and my family and friends keep telling me I'm not at all bad at it, even though I myself see many flaws in what I create and that I need to learn and improve.
*My opinion:*
I see art as product driven. 'Process driven', as you clearly show in your explanations, is just another term for 'money driven'. Furthermore, I can understand a true process driven argument in this discussion. But even then I argue that AI also goes through a process, just a little faster and perhaps a bit different process than the process a human uses to create art as biology (human) ain't no binary logic (computer). However, is the fact that an AI is faster than a human a reason to disqualify it as true art? If so, does that mean that the work of an experienced artist, who can create faster than a novice, is less 'art' as well? That's just ridiculous
No computer will ever pour it's heart and soul in to a piece. They don't have those things.
Most interesting commentary made here, very good insight and conclusions made about this controversy.
In my opinion, from past experience where I personally went through 4 years of commercial art- illustration career training back in the late 1960's and early 1970's using traditional media for creation, technological change has created new opportunities by destroying older forms. I recall back in the 1960’s and 1970’s how illustrations and advertising art were created using tools that are today to a large degree obsolete. These older tools took much more time learning and practice to use effectively. Such things like rub on graphic letters, color separation on transparent mylar type sheets, and other tools for print ready products. The use of computers and software made most of these old techniques and job skills obsolete.
The problem with this change is that the human mind and ability to keep up is not always there. There are some things that one can easily transition to and from yet others that are not within an individual’s abilities to migrate. We all have things we are good at and not too good at. You need a lot of luck, kind of like beating the odds at winning the lottery to keep a stream of success continuing through technological change that displaces long learned valuable skills. Looking at past precedent and historical evidence throughout time, most generations lived in an era where a set of job skills would last a lifetime and were valuable for survival and making a living, change came slow and more easily comprehended. The present generation on the other hand has seen too many changes in one lifetime that has totally disrupted valuable job skills to make a living and put humans into the shackles of constantly spending valuable time and money keeping up with the rat race of accelerated change. Hard earned learning skills and education have a much shorter life cycle of use in today's fast changing technological pace of moving forward. This pace will continue to accelerate as a necessary means of survival as the world's population continues to grow exponentially and a means of managing the growth is crucial.
In all fairness a lot of changes do provide convenience and make things easier yet at the same time de-values any worth for some time, effort and dedication towards the acquiring of any skills. Honestly, has all this technological change really improved our lives in a measurable degree that makes all the efforts to keep up with the rat race worthwhile? Bottom line being a worthwhile investment. Would the other possible outcome be more in line with the reality of what all this change is doing to our human race? Are we simply making things more complicated and mentally fatiguing as it destroys our ability to make a living and forces one to constantly learn new skills that are like rolling the dice?
Is there a possible connection with how all this technological change is destroying human morale and leading to more hopelessness and loss of confidence in one's ability to make a living from personally developed skills that are valuable in the marketplace? Seems the rate of worldwide protest in opposition to different things is growing increasingly in this generation along with an increase in crime and corruption within most countries and people roaming around the world being displaced out of desperation.
In all due fairness AI was not the biggest competition and challenge to creating art from scratch using traditional mediums or digital painting and drawing. The same thing happened over a century ago when photography became available to the consumer. Before photography many artists found work in portraiture since that was the only way to create a likeness of any visualization of a human as a means of memorabilia. Likewise, yet to a lesser degree landscape paintings, and still life. Photography won a large percentage in competition by replacing traditional medium art of drawing and painting for freelance and other established business means of making a living.
14:57 If you think "Unfair capitalization (basically, exploitation)" is the problem, the real problem is actually capitalism.
Exploitation isn't a bug it's a feature
Thanks for this balanced video that expands on the valid points of the artists as well as the developers.
6:18 is a nice summary of the whole situation!
I've actually seen both sides of the arguments _within_ the artist community. Paradoxically, it's usually the _product-driven_ artists that were the most upset. The value of art as a process is a also an emotional one, and ultimately what drives an artist forward. The product is the destination and does also have layers of value and meaning beyond just its monetary aspect. But that actually goes for AI-generated art and the atmosphere it creates by capturing the unique nature of a particular style. There are few words that have such a contested definition as the word "art". ("Love", perhaps) I can see why there is such a huge debate about whether "AI Art" meets some of these definitions, like an author's self-expression that's not given by something that's generated by an algorithm. Yet there is also a common notion that there is no actually original art, and what differentiates plagiarism from professional art is knowing how to steal from many different sources. In fact, some of the same artists that now talk about AI just stealing art argued about precisely that. (As the book titled "Steal Like an Artist" suggests, which I actually found through a TH-cam artist recommendation)
And as you mentioned at 10:15, plagiarism is about _the person_ falsely claiming work as their own. Whether that be an actual artist, or - In the case of AI art - the owners behind the technology. Don't demonize the tool - blame the user.
In that regard, AI basically taps into more sources of artworks as reference to create an images than any real person ever could. Of course there is a more unique conversion and perception when an actual person converts their reference material. But the technology is not actually taking any one artwork and selling it as theirs.
That said, using the images without consent _is_ an issue, as you pointed out at 9:05.
Thankfully, opt-out options are at least popping up that allow artists from excluding their art. But while Opt-out could still be considered a crummy approach (and laws in the future might even make this outright illegal) that ultimately won't prevent AI art to reach a level where it can meet the vast majority of art commissions.
I think the real issue is the underlying fear through the sudden shift in the art market, as noted by you at 11:03. Suddenly, technology made it possible to create art styles that previously required hours of work - let alone years of practice to hone that skill. So I can see why full-time artists are so upset right now. Particularly the less professional ones (as there is no target group left for them, losing to both cheap and clean-looking AI art that the average joe is content with, and not being able to keep up with the pros providing refined and expensive hand-crafted art for the connoisseurs.)
They now meet the fate that many jobs were previously done manually and lost to automation. Art was considered this bastion that computers and automation were for the longest time impossible to touch, but now suddenly reached a dumbfounding level of proficiency in the blink of an eye. Making a living as an artist now definitely got harder. But in magical wonderland where money is not a issue and merely becoming a good artist and enjoying the process is the goal, the situation actually got a lot better.
Because ironically, It's never been easier to create reference material to practice drawing specific objects, compositions, drawing styles or color palettes than before.
But as you implied at 11:16, this is a blessing for the society at large by making art products far more accessible than ever before.
For the consumer.
For the hobbyist painter.
And even for the _nonvisual artist_ that previously had no budget for enhancing their work of writing or music with pictures, this is a win/win situation.
Thank you for watching everything and giving such detailed responses to some of my points! I'm glad this has spurred more discussions like these. I've also been recommended "Steal Like An Artist" and as an artist, I actually believe the word "stealing" in art has got a lot more negative connotation than it should have. As humans, we learn by imitating and copying and there should be no difference between AI doing that and us doing that.
@@cynzyyy Thank you for providing the basis for these discussions!
You think "stealing art" has a more negative connotation within the art community?
So far my impression has been the opposite:
That the idea of "good stealing" (as opposted to plagiarism) is widespread in the artist community, mostly regarding "stealing" as a good habit or necessity even; for precisely the reasons you mentioned.
Whereas the average consumer is much more likely to look down on taking elements of other people's art, often even due to the misconception that a 'talented' person can simply draw from memory, and using references is cheating.
Either way, it probably explains why the louder parts of social media largely speak out against AI.
"AI can never replace what human artists can bring" - bold statement. Skin bags are so fond of themselves.
Ha ha, imagine a day when an AI will comment something like this.
I think this is a similar disruption to what record players were to musicians around the 30s. It doesn't fully replace live performance, but there are kinds of demand that can be satisfied with this mass produced art-like thing. Of course, for someone who values the process will not be satisfied with an AI-generated picture. For the rest, AI will outcompete artists by being cheaper.
I don't believe it's plagiarizing to AI generate "art" in the style of a human artist. You could just as well pay an artist to copy the style of another artist, and it's still a separate creation. The only difference is that now it's cheaper. Of course it's nice to specify if one takes a lot of inspiration from someone else, but it's not mandatory, because it's a new creation. Whether it's for profit or not shouldn't matter. Copyright protects the concrete, materialized creation, not the style.
Product driven honesty. In the end most people will see the end product rather then the process.
6:20 Answering your question about how I view art: Is it fair to say I see it both ways? Both as a process and a product.
My thought at this point is to mention that the law seems to agree with this view. AI "art" is a product, but it can not have copyright. Meaning it's a product that lacks the process. It can exist, but it isn't protected.But this is where, I think, the discussion takes a turn, because the artists, who's work is protected by copyright, says "we want to exclude our art from being a part of Ai training", and is that even reasonable? Copyright includes language about transformation because if a human artist uses another artist's work as inspiration, we accept that. But as soon as we turn that human experience into a process to produce a product, we want to be able to control it.
And, this is what _I_ think is what's at the core of the discussion. It's emotional. We think art, creativity, and the ability to create is a uniquely human thing. Then along comes a machine we've make something that turns that human experience into a text prompt with an output that matches anything a human could create. If that's possible, what does that mean we are as humans? What do we have left?
Okay, gonna watch the rest of the video now.
I think your solution to allow artists to only opt in is either short sighted or masking a desire to kill AI art. You repeatedly site an example where the prompt is "This in the style of a specific artists", but a lot of AI art prompts are more broad than that. "This in the style of a digital painting" or "Photo realistic" or "children's illustration", and in those cases it's still not a signle artist who's work is being is being infringed on, but everyone who's art can be described in that way.
Even in your example of a singular artist being parroted and their ability to opt into that, the most prolific artist in the world hasn't made enough work to begin to train an AI. If a single artist never published, say, a bowl of fruit, or an anime inspired mecha, then we can't use their dataset to make a Carmen Maranda mecha, because the dataset doesn't include examples the elements that would go into it. It's not just working with one hand tied behind it's back, it's working with both hands tied, a blind fold, and in a language you don't speak.
Imagine we decide that we don't want human artists to copy each other. Well, if you decide you want to be an artist, we'd first have to figure out if you've already seen anyone's art every, at any time. If so, you can't be an artists. But now we have a problem that no one can be an artist. However, since artists are obviously important to a thriving society, so we decide that we're going to take a certain portion of the population, kidnap them from their parents, raise them in rooms with plain white walls, and hope that they'd magically become the next Monet. Oh, except Monet trained under and copied artists of his day, so that's never going to work. At best we're going to send ourselves back to cave paintings. Cave paintings of rooms with white walls.
I hope my hyperbolic example makes my point. Trying to control and limit the training data doesn't simply limit what's possible. It makes it impossible. And if that's what you want, then just say that. Be honest. You don't like AI art and you want it to go away. That's fine, if that's the position you want to take.
As I said before, I feel like AI art threatens what it means to be human. We're taking something that we thought was uniquely human and this inhuman machine is producing it with comparable quality. That scares us to our core.
But there is hope. Because in reality there's nothing that a machine learning process can do that hasn't already been done. Sure, it can remix in ways that looks like creativity, and in a lot of ways remixing _is_ creativity, but it can't create anything new. The frontiers of imagination are still the domain of humans.
We've seen this before. When photography developed to the point of threatening portrait artists jobs, yes, some artists lost their jobs, but other discovered new modes of expression. Cubism, surrealism, abstract, art deco, and a whole host of new artists expressions were invented. Meanwhile, people with cameras were perfecting their craft with amazing framing, double exposure, and other tricks. This thing that we though would destroy art actually expanded our ability to create.
I think AI will do the same. It's not a destruction of the process, it's the evolution of it.
Unfortunately this won't do much to satisfy the emotional reaction of those who feel personally threatened or slighted by machine learning, and I get it.
I guess I would recast your argument as not a fight between Product and Process and a fight between Reason and Emotion. The Emotion says "You're stealing our process" and the Reason says "It's just a Product". But more than that Emotion will pound on anything they can to get attention, and Reason, I'm afraid, is gonna just ignore them and move on.
This video kind of gets to the problem but it doesn't address it: It is possible to make a fair system to create the databases from which the algorithms will learn.
AI companies are processing artist's work in an unfair and even illegal manner, and these companies need to be held accountable for this infringements.
An not perfect example on how to do it correctly is how music generator algorithms treats musical copyrighted material
Great efforts in contributing into the discourse. I think one bit that's lacking in the whole discourse of AI/Art/Design/Visual creation is notion of monetising "authenticity" instead of "time"/"emotion of the artist". What makes one type of art more authentic than the other?
AI allows us to shift towards producing ideas faster. Instead of the reliance on craft/"process" driven. I think it'll also be very interesting to bring in perspectives of generative artists.
Also, that's not how copyright works. It's not plagiarism if you create something in someones style - a style can not be copyrighted, only individual works. The copyright infringement claim by AI is mostly about the files that it generates and stores from the images it analyzed.
I'm super info long videos. I'm a podcast enjoyer after all. Keep it up! And thank you for covering this topic.
I agree with almost everything you said, but I do think that the onus should be on these companies to track the images they use. The datasets should *all* be publicly available, the artists should be able to find their artworks in these sets, and they should be able to opt-in/out. There needs to be more responsibility on the part not only of the companies providing the AI art producers, but the people who put together these datasets. Right now, it's all "I don't know, I just used it, I just found the image and included it" etc. But that's really not good enough for something that will cause such a profound change in society, in people's individual lives and livelihoods, in the accessibility/valuation of art. These people and companies need to get themselves in gear - they are mostly individuals or groups of computer nerds, not ready to deal with these kinds of changes that they have brought about. I respect that, and I also respect that it will take a lot of time and money to implement my idea. But I think that's more important. AI is supposed to be efficient, but it should ultimately be a *choice*. I should be able to choose whether I go to a human artist or computer. I should be able to choose whether I want my art/photo/likeness included in a dataset. I should not be told that it's too much time and money and, therefore, that that choice is a waste of time and money. These are things that should've been thought about long before this point, ideally. But they weren't. So we need to do it now, while it's happening. And if it messes up the AI revolution for a few years, I think that's acceptable for humans to be properly respected AND responsible.
Alright why would 24/7 guy care that artist gets replaced? He dosent really care that it is made by a Human or a machine. If he likes it, he will buy it.
It is way more easier just simply to watch the definition of art, and everything will be clear: "the expression or application of human creative skill and imagination, typically in a visual form such as painting or sculpture, producing works to be appreciated primarily for their beauty or emotional power." - Oxford Languages
Lower on the list of definitions (but currently rising) "decorative or illustrative elements in printed matter"
I don't normally comment on YT but I just found your channel today and I've been binging all your vids! Thank you for such beautifully presented and thoughtful content.
As a concept artist working in AAA games for my work, the ethics of AI really does concern me and I do think we should be questioning the morality behind it and where to draw the line. I really appreciate your research and insight into this.
thats wild!!!!! I was just thinking about how works are/content is created, before watching this video, and settled on the mantra "the product is the process, and the process is the product." I wonder if this conception of human-created works will be discussed later in the video
anyway he stole a urinal and wrote someone elses name on it so that part of the conversation been closed
It's not just a tool, it's a replacement, let's not just talking about business and all that, using AI to create something for urself, maybe just hanging on ur wall or just making it for personal use wallpaper was also a threat to any artist out there, it is stealing too doesn't matter is it for personal use or commercial use, only the artist themselves were allowed to use their own hard work to train a AI model for themselves, not anybody else that was not given permission to
people was actually commissioning AI to create art, not 'using' it
let me repeat, it's never just a tool, but a replacement, u can already tell by how many lesser famous artist's career was negatively impacted by it, there were plenty of tutorial out there teaching u how to use the 'lesser famous' artist's works to earn money, just one or two stealing jobs from that lesser-known artist was enough to kill his whole career
and thats a good thing
@@xinit69420 thats why ppl calling someone exactly like u trash human lol
AI art will not kill artists.
But it will replace mediocre artists for all the common work (minor graphic design, industrial drawings, cheap commercials/logo etc..;)
Artists will still exist but there will be no room for "average" artists, they won't have a job opportunity any longer.
Just like how industrialisation displaced artisans work. Artisans still exist, but only for high end products.
20 minutes I my ideal length of essay video. More is welcome.
This is the best articulation of the nuance of AI art. I have adopted this perspective in it's entirety, it's just really good.
Thank you so much!! Good to know that you don't hate it being too long haha
Thank you, Cynthia! Another great video!
Here is my take on this:
In the end, the questions are: Who are we? What do we value? What makes us human? How are we different from non-humans, from machines, or even from each other? What makes us unique, both as humans and as individuals?
My understanding is that whatever we are, what humans want is to connect to other humans, in various ways, art being one of them. (Seen with Darwin’s eyes, ultimately we connect as a form of expressing our phenotype to find someone to mate with, and a community that helps ensure our survival and the survival of offspring. Our brains are set up to want and do this. A tangent point is that we could of course share art with animals, as part of our community, so it is not just about humans. But all of this only sets the stage for what living beings do, so let’s move on to see what happens on this stage.)
So, art is by necessity a form of human expression. We care about the people who create the art, about the provenance of art, and what thoughts went into creating art. - Even if we have to find our own interpretations and use art as an object onto which we are projecting our own thoughts, it is still about our human interpretations and projections, and how they relate to the art and its creators.
An interesting thought is: if an AI would not only create art by closely following specific human direction, but if it would create many pieces of art by following an automated process, would we still be interested in every one of the many pieces of art it “creates”? - My answer is: no, we wouldn’t.
Because all this “art” that the automated AI generates would just be the product of some process within the automated AI. And unless this “art” connects us to something of value to us, which ultimately are other humans, life, community, we have no reason to care.
A slightly different question: what if someone selects “interesting” pieces of “art” from the stream of creations of that automated AI? Would we be interested in those? - My answer is: yes!
Because in this scenario, there is a selection process going on, and the human who selects the pieces of art and the reasons why they selected them are interesting to us again. The selected pieces of art connect us to another human, life, a community. The AI is only a tool used in the process of this creation.
Now, as we value human-made art, even if AI is used in its creation, we also want to value the human creator, and their work. And we recognize rights that come with the creation of art, such as copyright.
So, in the end, the core problem with art that has been created with the help of AI is that it makes the imitation of previously created art so much easier. We need to see how well copyright and its tests can handle this. The usual questions are: Is it a copy or a transformative work? And by how much? - And how do we handle the mass-creation of imitative works reasonably? Is the Millennium Copyright Act sufficient? What can we learn from similar problems on platforms like youtube?
I do believe that AI-generated art gives this problem a new dimension, because we probably need to define the boundaries between a copy, an imitation and a transformative work better. And what is “fair use” in this scenario? And we need to learn how to deal with these problems at a much bigger scale.
But fundamentally, we do have a mental framework that we can apply to deal with AI-generated art. We just need to learn how to apply this framework to the new challenges.
Oh, and how do we do that? Ok, I believe we need platforms that help with copyright tracking, consent arbitration and payments. So, what youtube does, just better. And partially automated, but with ways for humans to take part in every step, such as discovery, arbitration, negotiation, etc. And btw., another form of AI can help automate this.
This was a wonderful video! I feel you respectfully explained both sides
what someone said that stuck withme about this is that in the 1960s, they thought machines would free us from drudgery and let us do things we enjoy (ignoring the issue of companies just choosing, or being forced by economics, to simply discard workforces, and the difficulty of retraining), now the machines are here, they write code and draw pictures and write papers. Great, so if machines fill in for the drugery, and machines fill in for the art, what's left? I really think this is a situation in which the genie is out of the bottle and we're going to be feeling the bullshit coming off chatGPT for years to come.
I think something to acknowledge about "Misunderstanding 1" is that it's not just the user being a 'garbage human being.' They're just using the technology and if they ask it to make a portrait of themselves in the style of X artist and then want to use that on their social media - that's murky territory. The tool itself needs to have safeguards in place, and that falls on the companies making these tools to not just rush things out for hype, but to actually consider how they're implementing this tech. A simple example is GoogleMLM (a text to music AI tool), if you say, "make me a rock guitar lead line in the style of AC DC," it WILL NOT do it. If you use your own words to try to explain to GoogleMLM Angus Young's general guitar style (not an easy thing to do) then it will. This is a really basic level of protection for existing artists who are concerned about how their artwork is contributing to these tools, potentially to the detriment of their livelihood.
To be clear I'm hugely excited about AI in the art space. As a very mediocre musician, having AI tools to help me with the songwriting process, as well as mixing and mastering is huge.
Also for "Misunderstanding 3" the AI is not adding new experiences, it's just regurgitating what its seen based on user inputs. And regarding your question of 'is there a solution?' the AI is trained on a set of data. Companies should be responsible for the data they put in and it would not be unfeasible to create a sort of royalty structure similar to how music royalties (esp for streaming) work today. Lastly, your proposal couldn't work, right? I don't think most artists would want to outsource their work to AI, and also their own work probably wouldn't be a big enough dataset to run an AI image generator off of.
Again, I think AI is great, but when your gut tells you that for profit companies taking copyrighted artwork to train computers on, and individuals developing their own style to go on to make their own art and sell for profit... it just feels different.
2:45 Art "styles" are not actually copyrighted, and almost all artists will have art style similar to some other artist
If given a choice, most artists would rather spend 20+ hours on a piece than have AI generate it [Unless that artist wants maximum profit or can make the same living generating images]. Otherwise AI art already has a stigma attached to it.
There are many tools that allow artists to be more efficient and accurate. Like the camera lucida and camera obscura. But -
because somewhere in the 1900s someone labeled these as "cheating" no no one uses them. "Leonardo would never have used a camera obscura. Leonardo is amazing and Obscura is cheating and cheating bad! Leonardo good!" - even though Leonardo had access to Lorenzo de'Medici and the entire Medici family, including basically every single skilled glass maker, artist, scientist, and optometrist in Italy and was extremely curious so of course he would have had access to these devices.
Point is, once something in art turns sour, that creates a stigma that lasts hundreds of years where even if you create that thing, galleries, museums, and critics will shun you. AI art was here from day 1. The real question is, what is generative AI attempting to achieve when the same thing can and has been done far better with photobashing and composites?
The real AI tools for artists are things like Adobe Firefly. The rest of these generative garbage will be legislated into non-existence.
Excellent Video, I really like your AI Art proposal/solution.
The ideas conveyed in these videos are profound. They go to the root of the subject under discussion, Good stuff.
This discussion is an exact replication, of Taking a photo with a camera or drawing it, it's just a progress
So what happens when a bunch of artists try to claim a style when multiple people share a style? With the art world being as toxic and dog eat dog as it has become this problem will inevitably arise.
So glad I found your videos! The algorithm introduced me to one, and now I'm just finishing up watching almost all of them in one binge session. Well done! 非常好看!
only solution i can think of is... the redbubble model... you have databases and artists uploading to them, ONE DATABASE... for each AI... so the artist gets commission... percentage of how much was used goes to each artist...
One of the basic laws of economics is supply and demand. If the average person can create an image by just typing in a prompt, why will any company pay a trained artist to create images? Why would the general public think image making is a 'special' & 'skilled' activity, like they used to think in the past. The craft of image making could be relegated to a "bit of fun for youngsters" and not a viable career path for those who genuinely care about art and its rich history. A major part of the problem is perception by the lay person, as they could assume all images from now on are generated by 'just a prompt', even if they are not. I speak as a traditional artist who has seen the advent of digital and 3-D art and has had to retrain & embrace the changes to keep making a living over the years. Now this?
The S-U-B-S-C-R-I-B-E read gets me every time
loved loved loved ur video, and including the quotes from both parties, and I'm happy that u spoke of something that I do not see discussed enough... it's up for AI users to be aware of what this tool actually is, and use it responsibly, be it choosing to not use it at all, or at least be aware that it is based on the labour of all those who came before us, and how best it would be to use this tool to honor and respect their work instead of bashing all this and claiming it our own..
I really hope you keep posting content! I just found your channel and your do such interesting topics!
As an artist I can say that being scarred of AI is like believing in conspiracy theories, and again putting oneself in the mode of a 'victim' not a 'creator'. To be able to see AI as a creator I think it's amazing in this new era of these exciting new technologies to embrace this new vibe and mix it with one's creative process. Generative AI gives much cool ways to add to the process of developing of an idea. Rather than scrolling through Pinterest and get ideas from somebody's else art.
Also would like to highlight the part of comparing mass market vase to craft vase as art and not being a product... hmm there is a big difference between art vs craft, craft is product and has a particular purpose to use it in day to day life, when an art is something that has emotional or intelectual purpose to make us feel or think :)
I'd also add there is a phrase by someone "If they copy you it one more time confirms your genius". And it also comes from the creators mode, rather than from victim mode. It's like be above it all, keep creating and chasing your dreams...
as a tech enthusiast and hobbyist artist, this is quite a confusing situation to be in
I'm one of those evil "developers." I actually agree with you, and I contend that AI doesn't create art. It can't be original and it has only aesthetic appeal, no deeper meaning. That said, not all human-produced graphics have any deeper meaning or originality. Most art produced to use commercially doesn't rely on a genuine artistic process, either - it's made to satisfy a corporate checklist and earn an income.
All that said, even for art produced for commercial purposes, shouldn't be outsourced to AI. A human can make more meaningful connections and design decisions.
well big problem of AI art was that there is no any protection fence for human artist and no limit line for AI educating method. recently, U.S. court was judged that the copyright law will not agree of the artwork that made by non-human. This is will protect the leagal issue in a while, but it could be not the perfact answer. The big corporation and company are still supporting the AI developer for make cheap and useful AI to decrease the workers and employers. This mean, we can't limite the AI developing process with the large economic support. That's why law needs to chase the techonology development speed or go more faster than it to protect the people who are in the week protection environment. The law still needs to develop more to protect the people perfactly.
The fact that AI learns differently than humans is already a case for considering the way a database is used to be different of the inspiration atitude an artist has on works he admires and copy, and I mean by this that the whole point of the AI is that it 'learns faster' part. AI doesn't have acess to reality, it works as a logical system upon certain premises or axioms - humans do. We have to make sense of our experience to make sense of the world around, including other works of art that come to them as a given object as any other. When someone asks you to explain what a cat is, you'll be using the experiences you directly had with actual cats to do so, not a 'database' of second to third hand information - AI doesn't operate at that level, it only concludes logically a given data, which is the reason why those new technologies such as Nightshade are so able of fooling them. It's a computational mechanism, not a 'epistemical' one. Also, AI imitates a style not by it's true cause. Take for example an oil painting with loose strokes - an AI is able to imitate the manner without the access to a real brush or ink, and make it seems as if it were painted, when in fact it wasn't, is just a pixel analysis that resambles a picture taken from the original. If you know something about painting and already asked for an AI to make one in a particular style, you'll notice how strange normally it seems - they're normally too smooth, or its brush strokes are irrational. This is also true when it imitates a given trace, which is sourced on the mechanical predicates of our shoulders and wrists articulations, when it doesn't have any hand or arm whatsoever. It cannot considered as a mere 'copy and paste' copy, nor plagiarism of our usual kind, is in a level of mathematical imitation in various manners that human beings aren't capable of directly doing (if you take digital artists for example, which works involves pixels - though photographers might be included as well - it is as if the machine were able to copy their DNA/celular information there and transpose it, much more different than making an statue with clay). In my understanding is a copying machine by design and can't possibly stand in its own feet as it only produces second to third hand products (photographers and included in this), and in order to this be different (in any other regard as well) it needs much more than today computational capacities, it needs a way bridging the gap between its logical capacities and outside world.
Pues yo soy artista y creo que esa tecnologia es imparable, pero creo que nosotros los artistas tenemos que utilizar la IA como una extensión de la propia creatividad, empleándola para explorar nuevas ideas, o incluso como parte integral de sus obras finales. En este sentido, la IA puede ser vista como una herramienta que amplía las capacidades artísticas humanas, permitiendo la creación de cosas que antes eran impensables. al final el que hace la ai sigue siendo el creador, el que propone la idea, el que imagina las palabras para generar la imagen, eso cuenta como una expresion de si mismo
We give value to what we create.....time spent should not be considered but the message of the piece....what the artist intention of the piece.....most artist that is afraid is creating art....in a commercial way and sell art in a commercial.....honestly saying when i create something...its a part me that i dont others to have...but i have to set it Free....the value of it doenst lost of how much they pay for my work....but on how they see my value....THE art only devalued if ARTIST CREATOR chooses it to be so but not to the critics... art collectors....art lovers.....THE ONE YOU CAN HAVE MORE VALUE...IS A ART PIECE THAT AN ARTIST WILLINGLY AND SOULFULY GIVEN with no price......not in auction houses......Pricelessly given as a tokken of gift.....THATS all...i used A.i art.....
I'm on the Product side, Mostly because I like drawing manga and manga is more product than process.
Well put, another thing I thought would be a good addition was the pop up of financial accounts across all platforms promoting AI to make money, if you just search AI art, you see enough small financial channels grifting AI art to make money. This not only dampens the motivations of artists and illustrators such as myself to promote against them. One person can run multiple accounts to cast a wide net while keeping under the veil of anonymity but also benefit from the present state of AI art being able to produce art that can appeal to the general audience compared to one person having to manage a single account on multiple platform.
What AI art has also brought is this mindset of "artists being replaceable" now and that's a dangerous mindset to have, because of the grifters and promoters of these AI softwares repeats the same arguments, thus resulting in this oversimplification that people may not need to commission an artist anymore nor a company may not need to invest in 20+ artists/illustrators. This is a pervasive thought that has been shown true and continue to echo in similar form throughout these AI art debates in addition to the many you've already pointed which misses the entire arguments artists brings up to the people.
Not only does this dehumanizes artists even more than we already have been, it is an attack on art integrity. People often forget that the past two years, artists were already devalued and exploited by NFT grifters. Now we deal with AI grifters and it's worse because the backings from these are funded by many large corporations to develop these technology.
You got great points but your argument was focus on visual art which leaves out of the picture other branches of human creativity which is why other artists like music artists, writers etc should pay more attention to this issue also. I think we're all artists in a way or at least potential artists so maybe everyone should at least be aware of the issue.
Paying for the "process" is like paying 10x cost for a cookie because it was made with love. That just sounds like bad judgement to me. Fine art collectors pay exorbitant prices for art because its a concocted tax dodge. They they come up with post hoc bs reasons why its so precious and how art is so meaningful. No, they are just hustling the taxman. But I think people who don't know better hear that stuff and take it to heart.
I have to disagree with this notion that keeps getting passed around about AI being a "tool".
AI/ML Image Generators being cited as a "Tool for art" is extremely strange.
A tool is used to aid in the process of a product.
An AI/ML Generator literally just spits out an image.
That would be like a chef going, "Hey look at my new cooking tool." and pulling into the MCDs drive through.
@@birdedex Wouldn't that assume that marketing of commercial products has any impact on how the tech itself can be used? I fail to see the relevance in terms of the argument that it can be utilized as a tool.
Great video! The solution was already established by the same companies that created these models. It's an issue of willingness not capability. The music models already easily avoid training on music protected by record companies.
@Cynthia Zhou I disagree with your proposal because, first, I think "in the style of" should constitute copyright infringement. That artist's complaints are misguided because copyright infringement must occur at the time of sale, not at the time of production, that is to say that if I "sell art" than I as the "seller" should be legally responsible for selling a work which is infringing. Second, just like with the music industry, the act of dealing with an artist is both expensive and time consuming. You are asking people to trade 5 mins and $0 for returning to "pay me $1500 per image for commercial rights and I retain creative control," when in reality even if I like the "stroke style" of some artist maybe I don't actually like "THEIR ART WORK," only some specific brush stroke. In other words, it's a pointless loss of control and extra cost. I appreciate your argument that artists didn't get their piece and that there is nothing to learn from if no one makes anything new, that's completely right. I definitely don't think copyright actually applies to perception, in other words machine learning isn't infringing copyright because it is simply reading the data, that's actual nonsense. However, the truth of the matter is the company or companies took a bunch of art data, which they definitely required, and then sold the result. Talking about copyright actually derails the argument, because it's ignorant and only makes the arguer sound like an idiot to anyone who understands the technology, without exception. Your magic paint brush example was spot on, it hits the heart of the argument and the emotion of the argument while retaining reason and the appeal of the argument is the infringement of capitalist liberty, that of the artist in their contribution to the creation of that capital. I'm convinced enough by it that my stance is almost swayed, it would literally be swayed if I wasn't afraid to give up my civil liberties, i.e. the right to view, hear, or perceive copyrighted works in private as a human being, the right to own my own 'data' or in other words, the right to own my computer even if windows or osx is installed on it, and so on. If you let the legislators decide on too many of such things, I fear where we'll actually end up.
There is a way to solve this problem, if there was perceived value in collecting vast swathes of images at a time to train an AI model… then compensate the artists
Remember when NFT's where first starting out, and the first controversy was that they were stealing art, monetizing it, and giving no credit or compensation? I think that's what's going to happen on mass. Especially by less scrupulous actors, and people from countries in which it's difficult to litigate
There's value in taking vast amounts of photos of public spaces and streets. That doesn't mean car manufacturers, architects and storefront owners are owed money by google for their creations/designs being in street view.
I really love this video and co-sign all of the points you brought up here. As an artist seeing this debate flare up, I came to the same conclusion as you did, just using different words. I've spent my time creating "art," as in, what you all would expect, a piece of me that I created that was inspired and expressed. I've also spent my time creating "designs," that is, purely graphical things, usually through software, that people and organizations need. I don't see the latter as "art," but I get frustrated when "AI Art" is being touted as "art" when it's just "graphics," or as you described -- process. Neither is better than the other, this is just arguing semantics. What I don't like is all the fear mongering and finger pointing that's happening, especially coming from artists. I get embarrassed with how woefully uninformed they are about technology and are just saying it needs to be banned flat-out. That's not going to happen, technology will keep progressing. We need to determine how we perceive this. In the perfect world, as you described, we could train AI to our own works and produce things we simply don't have time for, just like outsourcing in-betweens from a Korean animation studio. If money was never exchanged, this would be fantastic. AI being used as a tool. But the issue is money is involved, artists are being displaced, and those that own the software charge per image. If this was really a tool, as it should be, then there would be no charge. No one charges me each time I hammer in a nail. Unfortunately, we don't live in that perfect world.
That said, plagiarism and stealing have ALWAYS been a problem in the art world -- we're just now seeing the tech bros give people license to do it through another degree of separation. But as you said, these bad actors would have done the same anyway. If NFTs, the metaverse, and crypto are anything to learn from, I would say that these grifters will be done in about a year and move onto the next big tech "disrupter." The general public will have a keener eye on how to detect AI art and will just brush it off as trivial digital bloat. Websites like DeviantArt, Pixiv, and other communities will establish guidelines and expectations when it comes to AI generated anything. Once all these scenarios eventually happen, artists will have a chance to take ownership of this tool and use it as it was intended.
You can both hate the grifters and thieves while still advocating for the advancement of advanced computer generated graphics.
To Mrs. Zhou, and to all people in the comment section, I understand the concerns that AI brings about. At best, I am optimistic with the progression made on AI, but at the same time, I am very cautious about it at the same time. Given its advancement in recent years like OpenAI's ChatGPT, I am sure it is all impressive stuff, but what I have long spoken about, and what I do fully perpetuate, is that having AI replacing every single job in existence, is very reckless, and tends to cause more harm than good. And God forbid we ever tie an AI to a missile silo. That will be an absolute travesty, and it will be a massive indictment to self-destruction. That's why there should be red lines that should not be crossed when it comes to using artificial intelligence in general.
At best, only most areas could be replaced with AI, completely or partially, but with the likes of cooking, policy enforcement (looking directly at you Google, DeviantArt, and TH-cam), and certain other fields, are better off not being used for artificial intelligence. At best, with the likes of editing and creating source code, they can be an exception, but by all means, they can make creating source code more efficient, and the editor just needs to edit other areas of the code so they can be satisfactory in the given outcome.
In addition to the topic of policy enforcement like bots to content ID systems, which is what the likes of what TH-cam, Google, and DeviantArt had been using for years, is just as reckless. Because for a start, artificial intelligence in general, lacks a conscience, thus they cannot accurately make the best of moral decisions, especially when it comes to punishing people and taking down contents. That's why policy enforcement, and taking down content that violates the policies and guidelines, are better off being handed to people rather than machines.
As for AI Art Generation on the other hand, I also do firmly believe we need more restrictive measures on those systems, like making them far more expensive to use, while fairly giving the person the right to own the art themselves.
In conclusion, those things I propose could much better compromises, rather than totally disallowing them. I was talking to the ChatGPT AI about the general rising hatred and fear of artificial intelligences, and even itself, does agree with that I am saying. I will provide the answers it gave within a text file, with the additional steps to ensure the safety of society. Are you more than willing to do whatever it takes to take this all on board to be examined by government leaders within democratic systems? Thanks very much for taking the time to read.
You presented all artists as being against AI art and seeing art as a process rather than a product. As a professional artist and designer myself who works with many other artists in the industry, from my experience this viewpoint is a result of a sampling bias. It might seem that way at first glance, but in reality this only applies to the vocal minority. The silent majority who are happy to simply add these new AI tools to their toolkit aren't going to be nearly as vocal online about this as are those who are panicking that AI is going to take their job.
I loved your metaphor with cheap nice looking vases at the dollar store. This is a good way of looking at it, as it makes it clear that we should all collectively prefer to live in a world that features nice looking vases at the dollar store, than in the world that exclusively features expensive handcrafted vases.
In my opinion, having artists consent to their art being seen by an AI is no different from having them consent to their art being seen by anyone else. They consented for their art to be seen in general, the moment they published it. I know that AI has seen my art, and I'm quite happy to have contributed an infinitesimal amount to AI understanding art concepts in general, as I am to have any human be inspired by or learn by looking at my art.
Many artists don't understand how AI works, and are under the impression that AI is taking their art and remixing it with other art to create new art, while in reality AI is simply looking at it and learning concepts.
When I create art, I use various processes and tools and mediums, rather than being married to one particular process, medium or style. This means that I don't agree with the sentiment SamDoesArt expressed in this video of "combining all our life experiences with how we approach every single brushstroke". That's a politically charged statement, and I'm quite sure he isn't literally doing that, as then each of his brushstrokes would take days.
Rather brush stroking is like breathing or typing, more of a subconscious process. And puring and combining our life experiences instead applies to our intent, to where we guide the artwork towards as we work, to what we envision or discover that we want the artwork to become in a given context. AI does not replace any of this, and in fact beautifully integrates into helping us explore the space of possibilities orders of magnitude faster.
What an amazing way of explaining things. Completely agreed with you.
Very good take!
What an amazing video! Amazing edit which sincs with everything you presented.
This is the best explanation I've seen so far. Well done. I look forward to watching more of your work.
If the value of human-created art comes from the process, then human artists have nothing to fear because people will continue to purchase human art for that reason. The problem is, people (generally) won't continue to purchase human art for that reason, because they put more value results.
Now, let's say that "real art" is always process driven, and AI art is fake art. Great!! Where does this leave us? People using fake art instead of real art. 🤷♀
It really doesn't matter what we do with semantics, people will gravitate toward a perceived value.
Embrace this new technology instead of complaining. The future doesn't stop. You've been using AI features in Photoshop like CTRL + T or the clone stamp for years, and now you're invoking the noble meaning of art. Let's have less hypocrisy.