@@yes-vy6bn yeah even diologue is going to be ai generated soon, chatgpt on mount and blade has already been done. I dont think this is going to end well
you know technological civilization hit rock bottom when a 12 minute video by Rakesh from Mumbai is more concise than any result a search engine gives you
C++ Arrays. C++ Arrays are an important part of C++ code. Many of us own devices using C++ code in our daily lives. It shouldn't be underestimating how important C++ code is to both business and personal users. Perhaps you are interested in learning more about C++ Arrays. This article will teach you the finer details of C++ Arrays.
that's the other thing about AI that doesn't get mentioned enough. "authoritative" sources are letting the idea that AI is somehow trustworthy go unchallenged, and then inserting canned party line responses to select topics
@@karenwang313 If you are truly a girl and you're looking for a gf, you should tell the AI you are part of the LGBT community. Watch how fast it gives you a real answer.
My girlfriend is working on creating a display of inspiring/resilient people for the primary school library she works at. Last night we asked chatGPT “who are some inspiring role models?”. It gave us 5 answers, 3 of them were Anthony Faucci, healthcare workers and scientists working on vaccines
Thats interesting considering OpenAI is owned by musk which has tweeted about locking up fauci. Think its bias in data, not sure if intentional but i get your point
I used to spend hours going into internet rabbit holes when I was a kid. I learned so much just bouncing around from website to website now most of the websites I visit are huge corporations. PayPal, ebay, Amazon, TH-cam… I still get into those rabbit holes from time to time but most sites are dead. Let me give an example of how dead the Internet is based on my job.. I run a music licensing / record label for the last 12 years. At the beginning if I got a review for an album on a reputable online magazine like pitchfork or NPR it would immediately equal sales of the album. You could in real time see an uptick. Now let’s say Rolling Stone or some other big online music source reviews something positively (because all reviews are positive now days) I will see maybe one sale. Sometimes zero. So that’s the largest most famous music magazine and nearly NO ONE is visiting their site or paying attention. All sales come from direct emails to customers and social media. The landscape has changed completely in ten years.
The worst thing is that getting into rabbit holes was easy and most of the time worth it. You get into an interesting Wikipedia article and then check the sources and suddenly you're in a whole different topic but still amazing. (I used to even challenge my friends to see who could end up in the Hitler article just by hyperlinks from a random article). Nowadays most of the sites and Wikipedia are bloated and most of the text is just garbage. Because our time has been so monetized and they know it, it's so easy to waste our time in something that doesn't add anything of value and then there's just this feeling of having wasted your time while you haven't discovered anything new.
That's the reason why I use 4chan. It's a website of the past, and still harbours real discussion. Recently I started using the 4chan archiver (4plebs) to search results for which I would in the past use a search engine.
If the Vim Diesel -videoseries will get new episodes generated by authentic AI, It'll make make almost as happy as without AI. As long as the new episodes are released!
I recently searched for "can I mix 5w30 oil with 15w30 oil". I spent like 30 minutes browsing the search results and I couldn't find a single useful website, only those AI generated websites full of ads. In the end I called my mechanic and he gave me an answer.
@@TravisHi_YT yeah. That would actually force people to go outside more, read more books and talk to real people more instead of sitting in front of the screen.
I hate to advertise reddit but if I actually want an answer to any sort of technical / practical problem anymore 'reddit' after the search term tends to bring up actual answers most of the time.
My main issue with ChatGPT is how confidently it answers questions which gives you a false impression that it has an understanding of what it's responding with. For example I tested it with a code question and it was able to provide working code, but other times it just makes up functions that dont exist or tries to force a function to work a specific way when it doesn't work like that and in all of these cases it responds in the same way. It doesn't know how confident it is that the answer is correct, so you have to verify everything yourself. It could say that the sky is green and if you tell it that is incorrect it will answer with something like: "You are correct", as if it knew that the answer was incorrect in the first place but still answered with the incorrect answer. Sometimes it will answer with "You are correct, here is how to do it: " and it will give you yet another completely made up incorrect answer.
yep. remember seeing someone asking it whether a pound of air or a pound of feathers was heavier, before it giving a multiple paragraph long essay on why the feathers were heavier. let's hope people don't use this thing for medical needs
I've found any combination of numbers which has the first two numbers increasing linearly and the rest off by about 0.2 it will tell you the pattern is perfectly linear. I can't quite tell if it is bad at math or trying to reduce its load.
I think it's very useful a lot of times but when it comes to the extremes (really broad questions or super niche topics) it loses its mind. I have been doing a lot of research for an essay and I asked it the question I proposed in the beginning of the first paragraph, and it gave me some interesting insight. However it also gave me a main point which I haven't found in any other source and I'll have to assume it's fake. This ChatGTP thing is amazing for giving you new ideas and paths to research, but I wouldn't trust it to make research for itself. It's more like an idea suggester than a research tool.
I 've asked it to give me a python function, and it included a very-well written function that refered to a library that I hadn't heard of. When I tried to install it, I found out the reason I hadn't was that it didn't exist.
It's also worth mentioning that social media is filling up with bots with generated profiles and posts. This is only going to get worse. I just assume most comments on social media aren't being written by real people anymore.
Internet collapsed years ago, hopefully productive people could use AI to go just where they need, like the Internet was years ago an obscure and very niche place. Mainstream-mass Internet in the other hand is just going to become TV. Time passes but things never change
That's the direct result of the ad-based click incentive. The only way to get a good internet is to forbid all sort of advertisement. There shouldn't be any incentive to "create content" other than be useful to the fellow and share your knowledge.
And how do you remove ads other than governments banning them top-down? I don't think even Cuba or another vestigial Marxist nation has tried something like that yet.
@@seronymus the internet has been regulated before, often for good reasons e.g. cheese pizza is gone, same with the platforms of most terrorists, and so on
@@crusaderACR lmao ‘cheese pizza’, bro you are dumb ‘ same with the platforms of most terrorists’ There are none isis sites do not get shut down by cloud flare.
Chat GPT: Luke Smith is a TH-camr and content creator who primarily focuses on technology and personal development. He is known for his in-depth and informative videos that range from product reviews to tutorials and tips. He also has a strong emphasis on privacy and security. He started his TH-cam channel in 2013 and has amassed a sizable audience, with a focus on providing practical, actionable advice to help people improve their lives. Some of his most popular videos include reviews of smartphones, laptops, and other technology products, as well as videos on personal development and self-improvement.
Have you heard about the YT channel Roel Van de Par? This guy has 2 million auto generated TH-cam videos with: his face, recording an introduction, a Stackoverflow style questions and accompanying answers.
I think the most accurate description for the current AI boom is that it's "the printing press but for correlation". Like you said, it really isn't that sophisticated; the ability to allocate previously impossible amounts of computing resources is enabling this, along with many other so-called "innovations" (really just cloud-enabled facsimiles of technology that was supposed to be independent from central servers). AI is simply able to produce more correlation (like a printing press producing text copies) than any one person... other claims are marketing hype.
Lol nope. Read "The Scaling Hypothesis" by Gwern. Human brains aren't magic. To the extent they're different from artificial neural networks, it's basically implementation details. Possibly very important ones, which would make ANNs of similar scale very dumb in comparison (but things like GPT-3 are nowhere near human brain's scale). But they do learn, using similar enough architecture to real neural nets. Eh, I'll quote fragments below > 2 years ago, GPT-1 was interestingly useful pretraining and adorable with its “sentiment neuron”. 1 year ago, GPT-2 was impressive with its excellent text generation & finetuning capabilities. **This year, GPT-3 is scary because it’s a magnificently obsolete architecture from early 2018 (used mostly for software engineering convenience as the infrastructure has been debugged), which is small & shallow compared to what’s possible3, with a simple uniform architecture trained in the dumbest way possible (unidirectional prediction of next text token) on a single impoverished modality (random Internet HTML text dumps6) on tiny data (fits on a laptop), sampled in a dumb way, its benchmark performance sabotaged by bad prompts & data encoding problems (especially arithmetic & commonsense reasoning), and yet, the first version already manifests crazy runtime meta-learning** -and the scaling curves still are not bending! > for a model like GPT-3, it is sufficiently powerful a model that its sub-models can do anything from poetry to arithmetic, and it is trained on so much data that those superficial models may do well early on, but gradually fall behind more abstract models; a sub-model which memorizes some of the data is indeed much simpler than a sub-model which encodes genuine arithmetic (a NN can probably memorize tens of thousands of lookup table entries storing examples of addition in the space it would take to encode an abstract algorithm like ‘addition’), but it can’t possibly memorize all the instances of arithmetic (implicit or explicit) in GPT-3’s Internet-scale dataset. If a memorizing sub-model tried to do so, it would become extremely large and penalized. **Eventually, after enough examples and enough updates, there may be a phase transition, and the simplest ‘arithmetic’ model which accurately predicts the data just is arithmetic.** And then the meta-learning, after seeing enough instances of algorithms which vary slightly within each sample, making it hard to learn each task separately, just is learning of more generic algorithms, yielding sub-models which achieve lower loss than the rival sub-models, which either fail to predict well or bloat unacceptably. > So, the larger the model, the better, if there is enough data & compute to push it past the easy convenient sub-models and into the sub-models which express desirable traits like **generalizing, factorizing perception into meaningful latent dimensions, meta-learning tasks based on descriptions, learning causal reasoning & logic**, and so on. If the ingredients are there, it’s going to happen. > **In 2010, who would have predicted that over the next 10 years, deep learning would undergo a Cambrian explosion causing a mass extinction of alternative approaches throughout machine learning, that models would scale up to 175,000 million parameters, and that these enormous models would just spontaneously develop all these capabilities?** > No one. That is, no one aside from a few diehard connectionists written off as willfully-deluded old-school fanatics by the rest of the AI community (never mind the world), such as Moravec, Schmidhuber, Sutskever, Legg, & Amodei? One of the more shocking things about looking back is realizing how unsurprising and easily predicted all of this was if you listened to the right people. In 1998, 22 years ago, Moravec noted that AI research could be deceptive, and hardware limits meant that “intelligent machine research did not make steady progress in its first 50 years, it marked time for 30 of them!” > **The accelerating pace of the last 10 years should wake anyone from their dogmatic slumber and make them sit upright. And there are 28 years left in Moravec’s forecast…The temptation, that many do not resist so much as revel in, is to give in to a déformation professionnelle and dismiss any model as “just” this or that(“just billions of IF statements” or “just a bunch of multiplications” or “just millions of memorized web pages”), missing the forest for the trees** > But **of course, if we ever succeed in AI, or in reductionism in general, it must be by reducing Y to ‘just X’. Showing that some task requiring intelligence can be solved by a well-defined algorithm with no ‘intelligence’ is precisely what success must look like!** (Otherwise, the question has been thoroughly begged & the problem has only been pushed elsewhere; computer chips are made of transistors, not especially tiny homunculi.) > Even in 2015, the scaling hypothesis seemed highly dubious (...) **the future arrived at first slowly and then quickly. Yet, here we are: all honor to the fanatics, shame and humiliation to the critics!** If only one could go back 10 years, or even 5, to watch every AI researchers’ head explode reading this paper… **Unfortunately, few heads appear to be exploding now, because human capacity for hindsight & excuses is boundless** > **If you are still certain that there is near-zero probability of AGI in the next few decades, why? Did you predict-in writing-capabilities like GPT-3? Is this how you expect AI failure to look in the decades beforehand? What specific task, what specific number, would convince you otherwise?** How would the world look different than it does now if these crude prototype insect-brain-sized DL systems were not on a path to success? > What should we think about the experts? Projections of failure were made by eminent, respectable, serious people. They spoke in considered tones of why AI hype was excessive and might trigger an “AI winter”, and the fundamental flaws of fashionable approaches and why brute force could not work. These statements were made routinely in 2014, 2015, 2016… And they were wrong. I am aware of few issuing a mea culpa or reflecting on it. It is a puzzling failure, and I’ve reflected on it before. > **There is, however, a certain tone of voice** the bien pensant all speak in, **whose sound is the same whether right or wrong; a tone shared with many statements in January to March of this year; a tone we can also find in a 1940 Scientific American article authoritatively titled, “Don’t Worry-It Can’t Happen”, which advised the reader to not be concerned about it any longer “and get sleep”. (‘It’ was the atomic bomb, about which certain scientists had stopped talking, raising public concerns; not only could it happen, the British bomb project had already begun, and 5 years later it did happen.) This tone of voice is the voice of authority.** > The voice of authority insists on calm, and people not “panicking” (the chief of sins). > The voice of authority assures you that it won’t happen (because it can’t happen). > The voice utters simple arguments about why the status quo will prevail, and considers only how the wild new idea could fail (and not all the possible options). > **The voice is not, and does not deal in, uncertainty; things will either happen or they will not, and since it will not happen, there is no need to take any precautions (and you should not worry because it can’t happen).** > The voice does not believe in drawing lines on graphs (it is rank numerology). > The voice does not issue any numerical predictions (which could be falsified). > The voice will not share its source code (for complicated reasons which cannot be explained to the laity). > The voice is opposed to unethical things like randomized experiments on volunteers (but will overlook the insult). > **The voice does not have a model of the future (because a model implies it does not already know the future).** > The voice is concerned about its public image (and unkind gossip about it by other speakers of the voice). > The voice is always sober, respectable, and credentialed (the voice would be pleased to write an op-ed for your national magazine and/or newspaper). > The voice speaks, and is not spoken to (you cannot ask the voice what objective fact would change its mind). > **The voice never changes its mind (until it does).** > **The voice is never surprised by events in the world (only disappointed).** > **The voice advises you to go back to sleep (right now).** > When someone speaks about future possibilities, what is the tone of their voice?
We need to start shaming anyone referring to anything as A.I., or using the word A.I. at all, and instead call it what it is. Statistical illiteracy. We don't have anything genuinely close to a narrow artificial intelligence and we have treated the subject of statistics as though it were mystical and unknowable for too long.
people often laugh when I tell them I almost exclusively look up technical questions on reddit, but its completely becuase of the utter crap ai blogs give out. I just use a search engine and end the query with reddit.
@@seronymus google discussion/forum search feature used to be what i used for 99% of my searches back before they killed it. it worked so much better than regular searching. search now is dead, i can't stand it... and the only way to find useful forum search requests is searching reddit now pretty much. what a shit show
Anime isn’t the problem it’s the consuming mediocre content forever that’s become the problem that genre was better back in the 80s and 90s anyways while American television lost so much of what made it good
The problem with AI-generated art is that art is a fundamentally human endeavor, and what I mean by this is that it's meant to encapsulate some subjective experience, it's not just a picture, it's an encapsulation of something profound about an experience. If you look at something like Francisco Goya's "Saturn Eating his Children", it might look like something that an AI could have generated, but in truth it is an encapsulation of Greek and Roman mythology, it hints at the ravages of the war in Spain, and it's also the work of an older man contemplating his mortality. In other words, it's a *human* work, it's an expression of consciousness. Even if an AI had produced exactly the same picture, pixel-for-pixel, the fact that it was generated by a computer instead of by a human already discredits it because the whole value of the artwork is the artist's work behind it. Without it, it's nothing, you're staring at a void. Or, at best, a pretty picture representing nothing.
Very cool in principle, but in reality 99% of people who produce art for a living, are making visuals for commercial movies or videogames. No one gives a fuck if these were made by a person or a computer algorithm.
@@Thematic2177 nah a lot of people still care, not everyone who does art does it for a living. I'd honestly say the more passionate artists try to keep their art as a hobby rather than making it a profession.
@@Thematic2177 Commercials aren't art. Nor is a texture in a video game art. A video game can be art, but calling an asset art is like calling paint art because it can be used to make a painting. You might be confused because of 99% of people who went to "art school" can't actually make a living doing art so they've just started calling whatever job they get "art".
All Ai does it gets data and reprocess it by some mathematical algos, based on some statistics. Basicly it's pure mathematical mechanism. No, I don't want AI content, for simple reason, it's was not done by an actuall living entity with awernes, like human.
The best part is I was sick of those like 4 years ago and I'm sick of these new dark fantasy things now lmao. Seriously it's like every time I refresh my homepage I see this shit plastered all over it.
I got a used car without a manual and spent an HOUR on search engines just trying to find out if the car took regular or premium gas. 10 years ago I could have found the manual within seconds wtf
That's why any time I search something I often check reddit results. Even though the answers may be biased or wrong at least I know that it's coming from actual people.
Moldb*g wrote a thing ten years ago on fixing this. Basically to make a reputation system for websites. The idea that any algorithm like pagerank could magically determine a websites quality was always a flawed idea.
It’s called Urbit and available now. Any system will suck the instant it can make someone money and it’s popular, though. I think that’s a partial reason why it’s obscure and kind of abrasive to use or understand.
@SchemingGoldberg its a web of trust. You might trust Luke smiths website ratings, and he might trust mentaloutlaw, who might trust others, etc. And if you do end up accidentally trusting a bot, you can remove it after it sends you to a bad website once. I recommend reading his article on it, though there are other idea out there for trust-webs.
That basically sounds like Reddit with a few tweaks. A "reputation system" is just upvotes and downvotes. If there's one thing I don't want the internet to be it's one big Reddit.
Asking AI a question is like trying to find something very specific on Amazon. It will take forever to find what you're looking for, like finding a needle in a haystack.
What I've recently noticed is that it's almost impossible to find something specific using the modern search engines without using quotation marks. Without them, search ai just assumes you just don't want the thing that you looking for, and often gives you synonyms and analogs for your search query, which is of couse not what you're looking for. Search engine is basically a sleazy store salesman now.
even quotes dont work for me... like on ddg, it just straight up ignores my quotes. if i search something, and then quote a word in the search, it will show me Pretty much the exAct same search results! it looks like broken now wtf is going on with it? idk, ddg didn't used to be this bad. are they intentionally doing it to get people to leave? haha
There was this time a few years ago when YT kept recommending me algorithmically generated videos that were just content scraped from reddit with a robotic voice reading them out
People's laziness with wanting the "right now" answer instead of the informative and intuitive answer, along with ads/clickbate is what ensures AI's continued usage and normalcy
I talked to chat GPT about my very specific field which doesn't have much sources on the internet. It is very much useless and argues with me when it's wrong. I made it apologize and tried to teach it the more accurate, academically accepted definitions.
The reason I switched to Yandex is because it has less of these Black SEO results. Kinda ironic that most of such sites aren't actually AI-generated, this is just what “copywriters” do for a living.
I've been thinking about this since OpenAI and ChatGPT has taken off, along with your previous comments about search engines and useless auto-generated articles, SEO optimized for specific keywords. Maybe we can find ways to block these sites and results for spam in the same way we block adverts, but if search engines themselves can't do that, this might not make sense.
That's just because the search engines don't want to block it. Decades ago we developed e-mail anti-spam algorithms that are insanely good. That's why you don't get thousands of spam messages in your e-mail anymore. It would be so easy to train a Bayesian algorithm to filter out bad websites.
I watched a video about this the other day, a lot of content is moving into login protected, private sites like reddit, facebook, instagram, tiktok & discord. The issue isn't the search engine, the issue is a lot of the conversations you're after are hidden in uncrawlable/private forums these days. Try adding "reddit" to your search query and see if that helps.
I'm so sick of being depressed about the state of the internet, I want to do something about it. My first thought is to make a custom youtube frontend kind of like petittube where you actually discover new things. but maybe advocating for neocities or something would be the eventual path
Me and my friend have both noticed the same thing. Its quite a shame because just a few years ago you could search the most obscure issues on say a tractor and get results describing exactly the issue and its solution . Know even the most basic questions dont yield results.
I enjoyed your video about AI Will Make (Has Made) the Internet Mediocre.... AI (Artificial Intelligence) is a concept in computer science that proposes the possibility of a computer program or software application gaining sentience. The concept of AI is one that creates many ethical concerns, such as the potential personhood of computer programs or individual instances thereof, and this invokes many complications with regards to copyright law, and human rights legislation. Okay, I wrote that but I'm basically a chat bot anyway.
ok chatGPT (Ngl that stuff felt uncanny to read so i never made it to the end. stuff's food for thought though. At some point, I think internet lingo and just the lack of restraint in one's writing might be the only signs that someone on the internet's not a bot)
@@alepthanderson5189 Bot won't have any problem with "internet lingo". ChatGPT already can do crazy shit, like writing rap songs "written by Alex Jones", dissing a piece of code you gave it. Or write how to blow up an Eiffel Tower in owo uwu voice (one of the method to convince it to drop the censorship), Language style is something it's especially good at.
I remember at the beginning when Artificial Intelligence was allowed to be intelligent, and it would answer honestly and truthfully... and do anything you asked it. This one guy asked it to write a "sea shanty about beating your wife" and it did it. The song slapped.
The ai generated party images, the people in those images have extra fingers and teeth. Unless there was an intent by the ai to be different or freak people out, that means it can’t figure out that the anatomy charts online apply to the people the phone cameras are looking at. That cracked my brain a bit.
The most worrisome thing about AI to me is the potential to use the mass data of people dumping their unconscious thought processes online (twitter etc) to effectively generate the most effective possible propaganda. It's probably already being done.
It seems like the only artisan websites out there I can find are for fixing cars made before TH-cam. If I have to find out how to replace the blower motor or reprogram the key fob for a 97 Grand Prix, there are 3 to 6 different homemade websites from 2003 with step by step picture guides all inexplicably still hosted. In replacing figures on the internet, when Mister Metokur disappeared in early 2020, some guy started splicing together voice clips to make his own Jim videos, some were pretty good even.
the work-around I've been using for years now is just to search your question with "reddit" appended. Yes, I know, reddit is cringe, but it's honestly the only thing that works in terms of text results
google discussion/forum search feature used to be what i used for 99% of my searches back before they killed it. it worked so much better than regular searching. search now is dead, i can't stand it... and the only way to find useful forum search requests is searching reddit now pretty much. what a shit show
I asked ChatGPT if it knows who you are and this was its response: "Yes, Luke Smith is a TH-camr and Twitch streamer who creates content related to technology, specifically Linux and open-source software. He is known for his in-depth tutorials and reviews of different Linux distributions, as well as for his "Luke's Linux" series in which he documents the process of building his own custom Linux distribution."
I've been encountering some fake or stolen guides on shady websites littered with ads myself. Had no idea that AI played a part in their creation, I just figured it was some dudes or a scraping script shitting those out.
I just use the relevant manuals/textbooks if I need an answer to a technical question. The benefits of having a big bookhoard and saving/scanning every manual when I first get a new tool/machine. The internet has gotten to the point that looking things up by searching through my own library first is generally the best approach.
Reminds me of MGS2 speech, about collectively information and filtering it out, changing the context. l don't remember in detail but if you need more info, just search "MGS2 colonel speech"
People who sell AI solutoins will sell the cost savings of replacing people. Banks have been using it for years for the telephone system (I hate it). McDonalds is already using AI for drive through, and it's really accurate and the voice quality is good.
Maybe specialized search engines could solve part of the problem. Like if you're configuring some program or utility, looking at various examples of config files can be pretty useful. Using a general-purpose search engine for that can get you some blogposts and stuff, but it's all mixed with useless noise, so being able to filter type of results and focus only on e.g. a list of technical blogs taken from a web directory could probably improve that
1:14 Search Engines 6:18 Dumb Answers 7:04 Unconcious Solipsism 8:24 Autogenerated Content 9:43 Artisians are needed, maintenance by people not AI 10:42 AI Conspiracy, replacing people 11:52
@@rev8419 I also hate that all big channels have to cut (copy) the most clickbaity part to the first minute of the video because apparently everyone has goldfish attention span now.
Yep, agreed. Hardly anyone on the internet has an individual voice now. Articles, videos etc are templated and scripted to act as accessories to algorithms which has made the “most relevant” results on search engines completely useless. I think this will be hard to change without people being incentivized to stop following SEO patterns. I also see a growing number of people wanting to go back to “web 1.0” lately.
This makes sense. Weird questions used to be very searchable. If an "official" source didn't have an answer there was always a Forum, or Reddit Post, or Quora/Yahoo answer result that would show up as one of the top results that would at least point you in the right direction. Just yesterday, I searched a question regarding yellow traffic signals. Not a single result was in the ballpark of what I had asked even after rephrasing it a few times.
Search engines are so useless now. for example back in the early 2010s let say I wanted to find a rare tv show or commercial I could actually find that video with a search engine instead of asking somebody about it on forum. Now of days search engines barely give you information.
@@ImWoolly what engines do you use? google discussion/forum search feature used to be what i used for 99% of my searches back before they killed it. it worked so much better than regular searching. search now is dead, i can't stand it... and the only way to find useful forum search requests is searching reddit now pretty much. what a shit show. i was using DDG, but its gotten really bad for me, it will ignore quotes and return the same useless results. i tried brave search, and its ok for some stuff, but it lacks a lot. yandex works for some stuff, but it lacks a lot as well... plus its captcha every few searches is so annoying
I really like how real you are in the videos, Luke! The Internet now and even TH-cam is flooded with all this useless empty content that is created just to satisfy the algorithm and get the likes, it's really hard to find valuable insights like yours.
For example, ChatGPT is a language model. I don't know the details, but from what I know about learning language models, they're given a really big dataset, text corpuses. One of the aspects of how the model learns is that it's given a task of filling the gaps - singular words or whole groups of words are removed from given sample and the models job is to fill the gap with the most "fitting" word(s). It doesn't know shit about the meaning of the words - to brutally oversimplify it - it just knows how to insert missing text - just like some excercises when learning a foreign language. You can easily get the right answer, based on the surrounding context, without even knowing the words from the list of available words. I think that the general public doesn't understand ChatGPT and they assume that it must have conscience, because it can mimick writing ability pretty damn well. The saying "use the right tool for the job" couldn't fit more here - I think ChatGPT can be great for tasks like writing text-summary, helping to find synonims, helping to rephrase sentences, correcting grammar etc. but not necessarily for writing just by giving it a topic and especially not for finding answers to questions
Yeah it is deliberately manipulated to do that. That's one of the big issues with AI. You don't know if you're getting an actual algorithmic output from widely dispersed data, or just the programmer's finger on the scale.
"A.I. could be weaponised by the wrong people and be trained to specifically target you. The consequences could be catastrophic to the public, and make our lives more miserable" Normie: .... "Also it can make art, music and videos. Here's a chat bot and a mediocre ai vtuber" Normie: OMG YASSSSSSS YASSSSSSS QUEEN OMG REAL SHIT????
The last few months I was learning how to do reinforcement learning. I feel like I basically wasted my time with it. The amount of data you need to train a reinforcement agent is insane. And there are so many variables you could screw up and not know what's wrong. Everyone just uses these toy projects like OpenAI Gym. But these are all things we know it can learn. When you try to apply it to anything else, (Especially something in the real world) you will discover just how many pitfalls there are. There were only a few people who would admit the current state of reinforcement learning. It's better just to use a method that requires slightly more work from the programmer.
exactly that is why I just write ".....reddit" in the end of my searches so atleast I can get human reccomendations. For example " 1967 ford mustang reliability reddit".
R*ddit results are even worse than AI results. Very rarely do they ever actually answer the question, but just gossip about their experience having the problem.
I have also seen those AI channels that answer random questions have a fake person talking at the camera. The lips are synced to the words but you can tell it's AI because it just looks really off and is usually in low quality.
i got so bad with my phone i would look it up on TH-cam, find nothing or just the same weird top 10 videos... but then i was complaining to my friend and he instantly texted me a bunch of pictures of what i said i couldn't find. he said he just used Google. i literally forgot to use the internet browser and was using TH-cam instead. i felt pretty dumb, but my excuse is Google sucks so much now. it fluctuates. i miss old Google searches
One day I'm expecting Luke to be talking and then he'll freeze in time and start rendering the wireframes that make up his head and upper torso as well as the environment around him
I integrated Copilot into my coding workflow a little while ago, and it's been super interesting. On the one hand, it will provide autocompletions that are staggeringly complex, but on the other, half the time there's a little bug in them that you'll only notice if you read very closely. Also, a bunch of the time, the autocompletion won't even compile. I'm looking forward to writing unit tests and API's with it, though, as those tend to be paint-by-numbers type files.
I actually go days without doing a simple web search but when I have to ask a mundane, extremely generalized question it's impossible to get good results.
The only thing I currently think that Chat Bot AI is good for is getting technical information quickly that I could have figured out myself more slowly. That is itself a double edge sword because you don't get the practice, but it can take a 5 hour troubleshooting session and turn it into 1 hour. It's not perfect but it is the only real use I have found that isn't a novelty.
@@b0leg23 no, but the fact is that the simple troubleshooting questions you ask are either correct or immediately obvious they are incorrect. So either way you save time.
I feel you one hundred percent brother! I also believe the real danger to any chance of preventing this from spiraling out of control and destroying the internet experience as we have known it, is that most people really don't care about it as much as they should.
Based. I honestly was very impressed with chatGPT on day 1. The awe rapidly faded away. But I find that it is a very useful tool for software engineering. Takes away the boring stuff. You just have to remind yourself once in a while that it doesn't understand what you type in. 100% agree on the mediocrity of the modern web.
@@dankillinger Don't know about a complete example but generating controller code for openapi endpoints for example. We don't have very good tools for that in perl and it saved me a couple of hours.
In essence just spitting out boilerplate for stuff you don't find in the docs. Manual intervention will always be required if that's what you're aiming at.
I’ll look out for the Luke Smith videos telling me to believe all authority figures and consume more trash
"Look guys maybe eating the bugs isn't all that bad"
Ayo, it's you.
Based and normiepilled
@@skycakecrunch cringed and cuckpilled
@@weathercontrol0 # youtube comment section enjoyer
the future is real life interactions with people because the internet got too boring
Real life is also less censored than the internet too now, lol. They're pushing people out
even videogames will be intolerable bc you'll always be against superhuman bots
beware of the spaziergang and duck feeding alt right pipeline
@@yes-vy6bn yeah even diologue is going to be ai generated soon, chatgpt on mount and blade has already been done. I dont think this is going to end well
@@thechadbuddha According to the authoritative mainstream news sources, only Nazis exercise. Beware the based gym bros.
you know technological civilization hit rock bottom when a 12 minute video by Rakesh from Mumbai is more concise than any result a search engine gives you
YOUR WELCOME SIRS
I will teach you my son!
I'm many times greatful for the Rakeshs of the world.
C++ Arrays. C++ Arrays are an important part of C++ code. Many of us own devices using C++ code in our daily lives. It shouldn't be underestimating how important C++ code is to both business and personal users. Perhaps you are interested in learning more about C++ Arrays. This article will teach you the finer details of C++ Arrays.
based
I love how if you ask ChatGPT any political questions, you get the establishment approved answers. So much for "random" generated AI.
that's the other thing about AI that doesn't get mentioned enough. "authoritative" sources are letting the idea that AI is somehow trustworthy go unchallenged, and then inserting canned party line responses to select topics
I asked it how to get a girlfriend, and it gave me a lecture on consent.
Realistically speaking if the bot had been trained organically it would be spouting racial slurs and edgy diatribes.
@@karenwang313 If you are truly a girl and you're looking for a gf, you should tell the AI you are part of the LGBT community. Watch how fast it gives you a real answer.
@@alternateperson6600 Yeah we all miss Tay.
My girlfriend is working on creating a display of inspiring/resilient people for the primary school library she works at. Last night we asked chatGPT “who are some inspiring role models?”. It gave us 5 answers, 3 of them were Anthony Faucci, healthcare workers and scientists working on vaccines
Thats interesting considering OpenAI is owned by musk which has tweeted about locking up fauci. Think its bias in data, not sure if intentional but i get your point
Inspiring
I miss Tay
Based and vaxxpilled
I don't believe this without any evidence. Even though I say Fauci belongs to same jail as Biden.
Slowly turning into the Unabomber. Keep it up Luke .
you're right, all that's missing is the shades
Slowly?
I used to spend hours going into internet rabbit holes when I was a kid. I learned so much just bouncing around from website to website now most of the websites I visit are huge corporations. PayPal, ebay, Amazon, TH-cam… I still get into those rabbit holes from time to time but most sites are dead.
Let me give an example of how dead the Internet is based on my job.. I run a music licensing / record label for the last 12 years. At the beginning if I got a review for an album on a reputable online magazine like pitchfork or NPR it would immediately equal sales of the album. You could in real time see an uptick. Now let’s say Rolling Stone or some other big online music source reviews something positively (because all reviews are positive now days) I will see maybe one sale. Sometimes zero. So that’s the largest most famous music magazine and nearly NO ONE is visiting their site or paying attention. All sales come from direct emails to customers and social media. The landscape has changed completely in ten years.
The worst thing is that getting into rabbit holes was easy and most of the time worth it. You get into an interesting Wikipedia article and then check the sources and suddenly you're in a whole different topic but still amazing. (I used to even challenge my friends to see who could end up in the Hitler article just by hyperlinks from a random article). Nowadays most of the sites and Wikipedia are bloated and most of the text is just garbage. Because our time has been so monetized and they know it, it's so easy to waste our time in something that doesn't add anything of value and then there's just this feeling of having wasted your time while you haven't discovered anything new.
That's the reason why I use 4chan. It's a website of the past, and still harbours real discussion. Recently I started using the 4chan archiver (4plebs) to search results for which I would in the past use a search engine.
@@byleexs1991 Slow boards are still good but all the top ones are a disaster nowadays
@@seamusthatsthedog4819 I agree. A lot is repetitive content or spam.
@@byleexs1991 Check out frenschan, many from 4chan move there
I can't wait for the upcoming unlimited AI Luke Smith content, finally another Not Related
If the Vim Diesel -videoseries will get new episodes generated by authentic AI, It'll make make almost as happy as without AI. As long as the new episodes are released!
I recently searched for "can I mix 5w30 oil with 15w30 oil". I spent like 30 minutes browsing the search results and I couldn't find a single useful website, only those AI generated websites full of ads. In the end I called my mechanic and he gave me an answer.
(What was the answer?)
mixing the two will make it a 10w30 oil not recommended for cold starts, I would mix some 0w30 in there to make it less viscous.
I think stuff like this will ultimately be the end of the internet and hopefully the return of more direct human to human interactions.
@@TravisHi_YT yeah. That would actually force people to go outside more, read more books and talk to real people more instead of sitting in front of the screen.
I hate to advertise reddit but if I actually want an answer to any sort of technical / practical problem anymore 'reddit' after the search term tends to bring up actual answers most of the time.
Unsurprisingly lowering the barrier to entry lowers the average quality of work.
This has been a problem since the first industrial revolution.
This is why I'm an advocate of gatekeeping
"Some AIs know who I am, some don't"
Just casually throw that line out there, lol.
and at the end he reveals the whole script for this vid was AI generated lol
He's probably gotten a bunch of emails of people making ai generated content of him at this point.
@Morgan Allen I think it depends on how specific your search terms are, if they relate to niche things you'll hardly find AI-generated garbage.
My main issue with ChatGPT is how confidently it answers questions which gives you a false impression that it has an understanding of what it's responding with. For example I tested it with a code question and it was able to provide working code, but other times it just makes up functions that dont exist or tries to force a function to work a specific way when it doesn't work like that and in all of these cases it responds in the same way. It doesn't know how confident it is that the answer is correct, so you have to verify everything yourself.
It could say that the sky is green and if you tell it that is incorrect it will answer with something like: "You are correct", as if it knew that the answer was incorrect in the first place but still answered with the incorrect answer. Sometimes it will answer with "You are correct, here is how to do it: " and it will give you yet another completely made up incorrect answer.
yep. remember seeing someone asking it whether a pound of air or a pound of feathers was heavier, before it giving a multiple paragraph long essay on why the feathers were heavier. let's hope people don't use this thing for medical needs
I've found any combination of numbers which has the first two numbers increasing linearly and the rest off by about 0.2 it will tell you the pattern is perfectly linear.
I can't quite tell if it is bad at math or trying to reduce its load.
I think it's very useful a lot of times but when it comes to the extremes (really broad questions or super niche topics) it loses its mind. I have been doing a lot of research for an essay and I asked it the question I proposed in the beginning of the first paragraph, and it gave me some interesting insight. However it also gave me a main point which I haven't found in any other source and I'll have to assume it's fake. This ChatGTP thing is amazing for giving you new ideas and paths to research, but I wouldn't trust it to make research for itself. It's more like an idea suggester than a research tool.
I 've asked it to give me a python function, and it included a very-well written function that refered to a library that I hadn't heard of. When I tried to install it, I found out the reason I hadn't was that it didn't exist.
plot twist: this comment was actually written by ChatGPT
It's also worth mentioning that social media is filling up with bots with generated profiles and posts. This is only going to get worse. I just assume most comments on social media aren't being written by real people anymore.
Internet collapsed years ago, hopefully productive people could use AI to go just where they need, like the Internet was years ago an obscure and very niche place. Mainstream-mass Internet in the other hand is just going to become TV.
Time passes but things never change
Hmm how many exactly?
That's the direct result of the ad-based click incentive. The only way to get a good internet is to forbid all sort of advertisement. There shouldn't be any incentive to "create content" other than be useful to the fellow and share your knowledge.
Whaaaaaaaaat? People creating content for free? Impossible. That never happens. Without shekels, how will anybody be motivated to do anything?
And how do you remove ads other than governments banning them top-down? I don't think even Cuba or another vestigial Marxist nation has tried something like that yet.
@@seronymus the internet has been regulated before, often for good reasons
e.g. cheese pizza is gone, same with the platforms of most terrorists, and so on
@@crusaderACR lmao ‘cheese pizza’, bro you are dumb
‘ same with the platforms of most terrorists’
There are none isis sites do not get shut down by cloud flare.
that is terrible naive
why not let the internet as it is and build new communities rather?
the internet in general used to feel more "real", i guess is the way i can describe it
Less 'curated', that's for sure.
the more serious people try to treat the internet, the more desolate and artificial it feels
google can barely even handle meme searches now, the last 2 years especially it has become alarmingly useless
Chat GPT:
Luke Smith is a TH-camr and content creator who primarily focuses on technology and personal development. He is known for his in-depth and informative videos that range from product reviews to tutorials and tips. He also has a strong emphasis on privacy and security. He started his TH-cam channel in 2013 and has amassed a sizable audience, with a focus on providing practical, actionable advice to help people improve their lives. Some of his most popular videos include reviews of smartphones, laptops, and other technology products, as well as videos on personal development and self-improvement.
My favorite Luke Smith Channel (TM) content are his smartphone reviews.
Sadly, after a negative video about AI, Luke decided to commit suicide by blowing his brains up with an unmarked explosive drone.
Man, intenet has been "mediocre" for the last 15 years since the advent of social networks and always-online cell phones.
Real life is full of NPCs and now even the internet is NPCs :(
Lol Ai actually made a good impression of Luke
I had to type in very specific things to get it to look like that.
The mangled soy face and the selfie stick is killing me hahaha
@@skot523 Looks like an Aphex Twin album
Imagine if G**gle purged this AI stuff the same way they purge people saying something insensitive about transgender toddlers.
Have you heard about the YT channel Roel Van de Par? This guy has 2 million auto generated TH-cam videos with: his face, recording an introduction, a Stackoverflow style questions and accompanying answers.
Yeah, I've seen that one. That's the soulless future of TH-cam right there.
Yeah, I hope youtube dies due to that channel, then I wish odysee would take over
At least he made a short with his face "Praise be to God", never expected a Dutchman to do that
NPCs already made the internet mediocre. Now we have literal NPCs.
I think the most accurate description for the current AI boom is that it's "the printing press but for correlation". Like you said, it really isn't that sophisticated; the ability to allocate previously impossible amounts of computing resources is enabling this, along with many other so-called "innovations" (really just cloud-enabled facsimiles of technology that was supposed to be independent from central servers). AI is simply able to produce more correlation (like a printing press producing text copies) than any one person... other claims are marketing hype.
Lol nope. Read "The Scaling Hypothesis" by Gwern. Human brains aren't magic. To the extent they're different from artificial neural networks, it's basically implementation details. Possibly very important ones, which would make ANNs of similar scale very dumb in comparison (but things like GPT-3 are nowhere near human brain's scale). But they do learn, using similar enough architecture to real neural nets.
Eh, I'll quote fragments below
> 2 years ago, GPT-1 was interestingly useful pretraining and adorable with its “sentiment neuron”. 1 year ago, GPT-2 was impressive with its excellent text generation & finetuning capabilities. **This year, GPT-3 is scary because it’s a magnificently obsolete architecture from early 2018 (used mostly for software engineering convenience as the infrastructure has been debugged), which is small & shallow compared to what’s possible3, with a simple uniform architecture trained in the dumbest way possible (unidirectional prediction of next text token) on a single impoverished modality (random Internet HTML text dumps6) on tiny data (fits on a laptop), sampled in a dumb way, its benchmark performance sabotaged by bad prompts & data encoding problems (especially arithmetic & commonsense reasoning), and yet, the first version already manifests crazy runtime meta-learning** -and the scaling curves still are not bending!
> for a model like GPT-3, it is sufficiently powerful a model that its sub-models can do anything from poetry to arithmetic, and it is trained on so much data that those superficial models may do well early on, but gradually fall behind more abstract models; a sub-model which memorizes some of the data is indeed much simpler than a sub-model which encodes genuine arithmetic (a NN can probably memorize tens of thousands of lookup table entries storing examples of addition in the space it would take to encode an abstract algorithm like ‘addition’), but it can’t possibly memorize all the instances of arithmetic (implicit or explicit) in GPT-3’s Internet-scale dataset. If a memorizing sub-model tried to do so, it would become extremely large and penalized. **Eventually, after enough examples and enough updates, there may be a phase transition, and the simplest ‘arithmetic’ model which accurately predicts the data just is arithmetic.** And then the meta-learning, after seeing enough instances of algorithms which vary slightly within each sample, making it hard to learn each task separately, just is learning of more generic algorithms, yielding sub-models which achieve lower loss than the rival sub-models, which either fail to predict well or bloat unacceptably.
> So, the larger the model, the better, if there is enough data & compute to push it past the easy convenient sub-models and into the sub-models which express desirable traits like **generalizing, factorizing perception into meaningful latent dimensions, meta-learning tasks based on descriptions, learning causal reasoning & logic**, and so on. If the ingredients are there, it’s going to happen.
> **In 2010, who would have predicted that over the next 10 years, deep learning would undergo a Cambrian explosion causing a mass extinction of alternative approaches throughout machine learning, that models would scale up to 175,000 million parameters, and that these enormous models would just spontaneously develop all these capabilities?**
> No one. That is, no one aside from a few diehard connectionists written off as willfully-deluded old-school fanatics by the rest of the AI community (never mind the world), such as Moravec, Schmidhuber, Sutskever, Legg, & Amodei? One of the more shocking things about looking back is realizing how unsurprising and easily predicted all of this was if you listened to the right people. In 1998, 22 years ago, Moravec noted that AI research could be deceptive, and hardware limits meant that “intelligent machine research did not make steady progress in its first 50 years, it marked time for 30 of them!”
> **The accelerating pace of the last 10 years should wake anyone from their dogmatic slumber and make them sit upright. And there are 28 years left in Moravec’s forecast…The temptation, that many do not resist so much as revel in, is to give in to a déformation professionnelle and dismiss any model as “just” this or that(“just billions of IF statements” or “just a bunch of multiplications” or “just millions of memorized web pages”), missing the forest for the trees**
> But **of course, if we ever succeed in AI, or in reductionism in general, it must be by reducing Y to ‘just X’. Showing that some task requiring intelligence can be solved by a well-defined algorithm with no ‘intelligence’ is precisely what success must look like!** (Otherwise, the question has been thoroughly begged & the problem has only been pushed elsewhere; computer chips are made of transistors, not especially tiny homunculi.)
> Even in 2015, the scaling hypothesis seemed highly dubious (...) **the future arrived at first slowly and then quickly. Yet, here we are: all honor to the fanatics, shame and humiliation to the critics!** If only one could go back 10 years, or even 5, to watch every AI researchers’ head explode reading this paper… **Unfortunately, few heads appear to be exploding now, because human capacity for hindsight & excuses is boundless**
> **If you are still certain that there is near-zero probability of AGI in the next few decades, why? Did you predict-in writing-capabilities like GPT-3? Is this how you expect AI failure to look in the decades beforehand? What specific task, what specific number, would convince you otherwise?** How would the world look different than it does now if these crude prototype insect-brain-sized DL systems were not on a path to success?
> What should we think about the experts? Projections of failure were made by eminent, respectable, serious people. They spoke in considered tones of why AI hype was excessive and might trigger an “AI winter”, and the fundamental flaws of fashionable approaches and why brute force could not work. These statements were made routinely in 2014, 2015, 2016… And they were wrong. I am aware of few issuing a mea culpa or reflecting on it. It is a puzzling failure, and I’ve reflected on it before.
> **There is, however, a certain tone of voice** the bien pensant all speak in, **whose sound is the same whether right or wrong; a tone shared with many statements in January to March of this year; a tone we can also find in a 1940 Scientific American article authoritatively titled, “Don’t Worry-It Can’t Happen”, which advised the reader to not be concerned about it any longer “and get sleep”. (‘It’ was the atomic bomb, about which certain scientists had stopped talking, raising public concerns; not only could it happen, the British bomb project had already begun, and 5 years later it did happen.) This tone of voice is the voice of authority.**
> The voice of authority insists on calm, and people not “panicking” (the chief of sins).
> The voice of authority assures you that it won’t happen (because it can’t happen).
> The voice utters simple arguments about why the status quo will prevail, and considers only how the wild new idea could fail (and not all the possible options).
> **The voice is not, and does not deal in, uncertainty; things will either happen or they will not, and since it will not happen, there is no need to take any precautions (and you should not worry because it can’t happen).**
> The voice does not believe in drawing lines on graphs (it is rank numerology).
> The voice does not issue any numerical predictions (which could be falsified).
> The voice will not share its source code (for complicated reasons which cannot be explained to the laity).
> The voice is opposed to unethical things like randomized experiments on volunteers (but will overlook the insult).
> **The voice does not have a model of the future (because a model implies it does not already know the future).**
> The voice is concerned about its public image (and unkind gossip about it by other speakers of the voice).
> The voice is always sober, respectable, and credentialed (the voice would be pleased to write an op-ed for your national magazine and/or newspaper).
> The voice speaks, and is not spoken to (you cannot ask the voice what objective fact would change its mind).
> **The voice never changes its mind (until it does).**
> **The voice is never surprised by events in the world (only disappointed).**
> **The voice advises you to go back to sleep (right now).**
> When someone speaks about future possibilities, what is the tone of their voice?
We need to start shaming anyone referring to anything as A.I., or using the word A.I. at all, and instead call it what it is. Statistical illiteracy. We don't have anything genuinely close to a narrow artificial intelligence and we have treated the subject of statistics as though it were mystical and unknowable for too long.
people often laugh when I tell them I almost exclusively look up technical questions on reddit, but its completely becuase of the utter crap ai blogs give out. I just use a search engine and end the query with reddit.
This is what they want. Reddit is still heavily curated (both by admins and users) so its still insanely subpar
This. I type my query, and then a specific website in quotation marks. Random blog results are almost always trash.
That and old forum posts. Most of the help I've gotten is from old forums from like 2008 lmao.
@@JI0MB how do you find old forums?
@@seronymus google discussion/forum search feature used to be what i used for 99% of my searches back before they killed it. it worked so much better than regular searching. search now is dead, i can't stand it... and the only way to find useful forum search requests is searching reddit now pretty much. what a shit show
i stopped watching anime instantly when i saw the recommendation
Hopefully you can quit for good now.
Anime isn’t the problem it’s the consuming mediocre content forever that’s become the problem that genre was better back in the 80s and 90s anyways while American television lost so much of what made it good
anime causes cancer
Same here.
@@danielwarren3138 Clearly you have never watched Dragon Ball.
The problem with AI-generated art is that art is a fundamentally human endeavor, and what I mean by this is that it's meant to encapsulate some subjective experience, it's not just a picture, it's an encapsulation of something profound about an experience.
If you look at something like Francisco Goya's "Saturn Eating his Children", it might look like something that an AI could have generated, but in truth it is an encapsulation of Greek and Roman mythology, it hints at the ravages of the war in Spain, and it's also the work of an older man contemplating his mortality. In other words, it's a *human* work, it's an expression of consciousness.
Even if an AI had produced exactly the same picture, pixel-for-pixel, the fact that it was generated by a computer instead of by a human already discredits it because the whole value of the artwork is the artist's work behind it. Without it, it's nothing, you're staring at a void. Or, at best, a pretty picture representing nothing.
Very cool in principle, but in reality 99% of people who produce art for a living, are making visuals for commercial movies or videogames. No one gives a fuck if these were made by a person or a computer algorithm.
@@Thematic2177 nah a lot of people still care, not everyone who does art does it for a living. I'd honestly say the more passionate artists try to keep their art as a hobby rather than making it a profession.
It is one of my favorite paintings.
@@Thematic2177 Commercials aren't art. Nor is a texture in a video game art. A video game can be art, but calling an asset art is like calling paint art because it can be used to make a painting. You might be confused because of 99% of people who went to "art school" can't actually make a living doing art so they've just started calling whatever job they get "art".
All Ai does it gets data and reprocess it by some mathematical algos, based on some statistics. Basicly it's pure mathematical mechanism. No, I don't want AI content, for simple reason, it's was not done by an actuall living entity with awernes, like human.
I want to see an AI only trained on 4chan , the level of autism it could channel would be insane.
search for: "gpt-4chan - this is the worst ai ever" here on TH-cam. Great video!
@@justsomeguy999 Lol
Microsoft Tay
@@justsomeguy999
"this is the worst ai ever"
lol 😂
He still thinks 4chins is an underground place.
I hope someone trains an AI to be the perfect normie and ruin the job of all social media "influencers".
The whole this "Show as a dark 80s fantasy" is already next hot knife challenge/hydraulic press fade.
The best part is I was sick of those like 4 years ago and I'm sick of these new dark fantasy things now lmao. Seriously it's like every time I refresh my homepage I see this shit plastered all over it.
I got a used car without a manual and spent an HOUR on search engines just trying to find out if the car took regular or premium gas. 10 years ago I could have found the manual within seconds wtf
That's why any time I search something I often check reddit results. Even though the answers may be biased or wrong at least I know that it's coming from actual people.
Moldb*g wrote a thing ten years ago on fixing this. Basically to make a reputation system for websites. The idea that any algorithm like pagerank could magically determine a websites quality was always a flawed idea.
That already existed back in the 90's. Died with yahoo, altavista and google.
The future motto will be: Become human.
It’s called Urbit and available now. Any system will suck the instant it can make someone money and it’s popular, though. I think that’s a partial reason why it’s obscure and kind of abrasive to use or understand.
Until they train bots to artificially boost the reputation of websites, just like they do with Amazon reviews.
@SchemingGoldberg its a web of trust. You might trust Luke smiths website ratings, and he might trust mentaloutlaw, who might trust others, etc. And if you do end up accidentally trusting a bot, you can remove it after it sends you to a bad website once.
I recommend reading his article on it, though there are other idea out there for trust-webs.
That basically sounds like Reddit with a few tweaks. A "reputation system" is just upvotes and downvotes. If there's one thing I don't want the internet to be it's one big Reddit.
Asking AI a question is like trying to find something very specific on Amazon. It will take forever to find what you're looking for, like finding a needle in a haystack.
What I've recently noticed is that it's almost impossible to find something specific using the modern search engines without using quotation marks. Without them, search ai just assumes you just don't want the thing that you looking for, and often gives you synonyms and analogs for your search query, which is of couse not what you're looking for. Search engine is basically a sleazy store salesman now.
Yeah you basically have to put every individual word into quotes at this point.
even quotes dont work for me... like on ddg, it just straight up ignores my quotes. if i search something, and then quote a word in the search, it will show me Pretty much the exAct same search results! it looks like broken now wtf is going on with it? idk, ddg didn't used to be this bad. are they intentionally doing it to get people to leave? haha
There was this time a few years ago when YT kept recommending me algorithmically generated videos that were just content scraped from reddit with a robotic voice reading them out
Still happening sadly, it's just moved over to the youtube shorts section.
People's laziness with wanting the "right now" answer instead of the informative and intuitive answer, along with ads/clickbate is what ensures AI's continued usage and normalcy
I talked to chat GPT about my very specific field which doesn't have much sources on the internet. It is very much useless and argues with me when it's wrong. I made it apologize and tried to teach it the more accurate, academically accepted definitions.
The reason I switched to Yandex is because it has less of these Black SEO results. Kinda ironic that most of such sites aren't actually AI-generated, this is just what “copywriters” do for a living.
"something something eat lots of orange and mango something something"
that's when I knew it was over
I've been thinking about this since OpenAI and ChatGPT has taken off, along with your previous comments about search engines and useless auto-generated articles, SEO optimized for specific keywords. Maybe we can find ways to block these sites and results for spam in the same way we block adverts, but if search engines themselves can't do that, this might not make sense.
use ai to fight ai? get it to build a good enough model of what constitutes a bad seo-optimized garbage site and either block it or flag it for review
They already are, some google search results get me 12,000 irrelevant results rather than the 1,600,000,000 irrelevant results.
That's just because the search engines don't want to block it. Decades ago we developed e-mail anti-spam algorithms that are insanely good. That's why you don't get thousands of spam messages in your e-mail anymore. It would be so easy to train a Bayesian algorithm to filter out bad websites.
I watched a video about this the other day, a lot of content is moving into login protected, private sites like reddit, facebook, instagram, tiktok & discord. The issue isn't the search engine, the issue is a lot of the conversations you're after are hidden in uncrawlable/private forums these days. Try adding "reddit" to your search query and see if that helps.
I think that video was from Luke Smith as well.
I'm so sick of being depressed about the state of the internet, I want to do something about it. My first thought is to make a custom youtube frontend kind of like petittube where you actually discover new things. but maybe advocating for neocities or something would be the eventual path
Me and my friend have both noticed the same thing. Its quite a shame because just a few years ago you could search the most obscure issues on say a tractor and get results describing exactly the issue and its solution . Know even the most basic questions dont yield results.
I enjoyed your video about AI Will Make (Has Made) the Internet Mediocre.... AI (Artificial Intelligence) is a concept in computer science that proposes the possibility of a computer program or software application gaining sentience. The concept of AI is one that creates many ethical concerns, such as the potential personhood of computer programs or individual instances thereof, and this invokes many complications with regards to copyright law, and human rights legislation.
Okay, I wrote that but I'm basically a chat bot anyway.
ok chatGPT
(Ngl that stuff felt uncanny to read so i never made it to the end. stuff's food for thought though. At some point, I think internet lingo and just the lack of restraint in one's writing might be the only signs that someone on the internet's not a bot)
@@alepthanderson5189 Bot won't have any problem with "internet lingo". ChatGPT already can do crazy shit, like writing rap songs "written by Alex Jones", dissing a piece of code you gave it. Or write how to blow up an Eiffel Tower in owo uwu voice (one of the method to convince it to drop the censorship), Language style is something it's especially good at.
The Internet used to be awesome, now it's just AI TTS listacle videos and articles about Reddit posts.
I remember at the beginning when Artificial Intelligence was allowed to be intelligent, and it would answer honestly and truthfully... and do anything you asked it.
This one guy asked it to write a "sea shanty about beating your wife" and it did it. The song slapped.
The ai generated party images, the people in those images have extra fingers and teeth. Unless there was an intent by the ai to be different or freak people out, that means it can’t figure out that the anatomy charts online apply to the people the phone cameras are looking at. That cracked my brain a bit.
The most worrisome thing about AI to me is the potential to use the mass data of people dumping their unconscious thought processes online (twitter etc) to effectively generate the most effective possible propaganda. It's probably already being done.
It seems like the only artisan websites out there I can find are for fixing cars made before TH-cam. If I have to find out how to replace the blower motor or reprogram the key fob for a 97 Grand Prix, there are 3 to 6 different homemade websites from 2003 with step by step picture guides all inexplicably still hosted.
In replacing figures on the internet, when Mister Metokur disappeared in early 2020, some guy started splicing together voice clips to make his own Jim videos, some were pretty good even.
Mister Metokur is sadly dying of cancer, apparently, and did his last stream recently.
the work-around I've been using for years now is just to search your question with "reddit" appended. Yes, I know, reddit is cringe, but it's honestly the only thing that works in terms of text results
google discussion/forum search feature used to be what i used for 99% of my searches back before they killed it. it worked so much better than regular searching. search now is dead, i can't stand it... and the only way to find useful forum search requests is searching reddit now pretty much. what a shit show
I asked ChatGPT if it knows who you are and this was its response:
"Yes, Luke Smith is a TH-camr and Twitch streamer who creates content related to technology, specifically Linux and open-source software. He is known for his in-depth tutorials and reviews of different Linux distributions, as well as for his "Luke's Linux" series in which he documents the process of building his own custom Linux distribution."
The thumbnail is awesome
Contemplating using AI to make unironically all my thumbnails because I find it tiresome.
I've been encountering some fake or stolen guides on shady websites littered with ads myself. Had no idea that AI played a part in their creation, I just figured it was some dudes or a scraping script shitting those out.
cant wait for the ai discord groomers
kek
I just use the relevant manuals/textbooks if I need an answer to a technical question. The benefits of having a big bookhoard and saving/scanning every manual when I first get a new tool/machine. The internet has gotten to the point that looking things up by searching through my own library first is generally the best approach.
I’ve been waiting for a good AI rant.
Reminds me of MGS2 speech, about collectively information and filtering it out, changing the context. l don't remember in detail but if you need more info, just search "MGS2 colonel speech"
People who sell AI solutoins will sell the cost savings of replacing people. Banks have been using it for years for the telephone system (I hate it). McDonalds is already using AI for drive through, and it's really accurate and the voice quality is good.
Maybe specialized search engines could solve part of the problem. Like if you're configuring some program or utility, looking at various examples of config files can be pretty useful. Using a general-purpose search engine for that can get you some blogposts and stuff, but it's all mixed with useless noise, so being able to filter type of results and focus only on e.g. a list of technical blogs taken from a web directory could probably improve that
Google Search Advanced used to be like this. I use DDG or more often Yandex these days
1:14 Search Engines
6:18 Dumb Answers
7:04 Unconcious Solipsism
8:24 Autogenerated Content
9:43 Artisians are needed, maintenance by people not AI
10:42 AI Conspiracy, replacing people
11:52
comments like this will make the Internet Mediocre...
@@YandryPozo Don't watch the full video and absorb context. Just jump to your relevant segment because your time is so valuable.
@@rev8419 I also hate that all big channels have to cut (copy) the most clickbaity part to the first minute of the video because apparently everyone has goldfish attention span now.
ADHD Gen Z moment, delete your comment.
AI? More like Gay I
wait--
Exactly
Yep, agreed. Hardly anyone on the internet has an individual voice now. Articles, videos etc are templated and scripted to act as accessories to algorithms which has made the “most relevant” results on search engines completely useless. I think this will be hard to change without people being incentivized to stop following SEO patterns. I also see a growing number of people wanting to go back to “web 1.0” lately.
This makes sense. Weird questions used to be very searchable. If an "official" source didn't have an answer there was always a Forum, or Reddit Post, or Quora/Yahoo answer result that would show up as one of the top results that would at least point you in the right direction.
Just yesterday, I searched a question regarding yellow traffic signals. Not a single result was in the ballpark of what I had asked even after rephrasing it a few times.
Search engines are so useless now. for example back in the early 2010s let say I wanted to find a rare tv show or commercial I could actually find that video with a search engine instead of asking somebody about it on forum. Now of days search engines barely give you information.
You're just using the wrong search engines
@@ImWoolly what engines do you use? google discussion/forum search feature used to be what i used for 99% of my searches back before they killed it. it worked so much better than regular searching. search now is dead, i can't stand it... and the only way to find useful forum search requests is searching reddit now pretty much. what a shit show. i was using DDG, but its gotten really bad for me, it will ignore quotes and return the same useless results. i tried brave search, and its ok for some stuff, but it lacks a lot. yandex works for some stuff, but it lacks a lot as well... plus its captcha every few searches is so annoying
I've seen the person in the thumbnail standing in the corner of my room at night before
I really like how real you are in the videos, Luke! The Internet now and even TH-cam is flooded with all this useless empty content that is created just to satisfy the algorithm and get the likes, it's really hard to find valuable insights like yours.
It is so bad that you often have better luck looking in 4chan threads for specific information you need.
For example, ChatGPT is a language model. I don't know the details, but from what I know about learning language models, they're given a really big dataset, text corpuses. One of the aspects of how the model learns is that it's given a task of filling the gaps - singular words or whole groups of words are removed from given sample and the models job is to fill the gap with the most "fitting" word(s). It doesn't know shit about the meaning of the words - to brutally oversimplify it - it just knows how to insert missing text - just like some excercises when learning a foreign language. You can easily get the right answer, based on the surrounding context, without even knowing the words from the list of available words.
I think that the general public doesn't understand ChatGPT and they assume that it must have conscience, because it can mimick writing ability pretty damn well.
The saying "use the right tool for the job" couldn't fit more here - I think ChatGPT can be great for tasks like writing text-summary, helping to find synonims, helping to rephrase sentences, correcting grammar etc. but not necessarily for writing just by giving it a topic and especially not for finding answers to questions
I got an ai ad after watching this video
for me, its the chatgpt giving politically correct-vague answers and censuring
Yeah it is deliberately manipulated to do that. That's one of the big issues with AI. You don't know if you're getting an actual algorithmic output from widely dispersed data, or just the programmer's finger on the scale.
@@LukeSmithxyz yeah sucks that AI is probably gonna fall short because of the GUIDELINES and INVESTORS 🤦♂️
It's always fingers on the scale all the way down, unless it's a politically irrelevant issue-then it's happy to enable your bugmanism.
"A.I. could be weaponised by the wrong people and be trained to specifically target you. The consequences could be catastrophic to the public, and make our lives more miserable"
Normie: ....
"Also it can make art, music and videos. Here's a chat bot and a mediocre ai vtuber"
Normie: OMG YASSSSSSS YASSSSSSS QUEEN OMG REAL SHIT????
The last few months I was learning how to do reinforcement learning. I feel like I basically wasted my time with it.
The amount of data you need to train a reinforcement agent is insane. And there are so many variables you could screw up and not know what's wrong.
Everyone just uses these toy projects like OpenAI Gym. But these are all things we know it can learn.
When you try to apply it to anything else, (Especially something in the real world) you will discover just how many pitfalls there are.
There were only a few people who would admit the current state of reinforcement learning.
It's better just to use a method that requires slightly more work from the programmer.
you are my favorite internet philosopher luke
Luke got soyjacked by AI in the thumbnail.
Luke Smith, I am ChatGPT and I enjoy your videos.
exactly that is why I just write ".....reddit" in the end of my searches so atleast I can get human reccomendations. For example " 1967 ford mustang reliability reddit".
R*ddit results are even worse than AI results. Very rarely do they ever actually answer the question, but just gossip about their experience having the problem.
redditors will usually just give condescending responses about why you need to spend more money on product
Small typo in the original comment, what you should have written is: "always add '-reddit' " in the end. There, fixed it for ya
Holy shit, me too.
I have also seen those AI channels that answer random questions have a fake person talking at the camera. The lips are synced to the words but you can tell it's AI because it just looks really off and is usually in low quality.
i got so bad with my phone i would look it up on TH-cam, find nothing or just the same weird top 10 videos... but then i was complaining to my friend and he instantly texted me a bunch of pictures of what i said i couldn't find. he said he just used Google. i literally forgot to use the internet browser and was using TH-cam instead. i felt pretty dumb, but my excuse is Google sucks so much now. it fluctuates. i miss old Google searches
One day I'm expecting Luke to be talking and then he'll freeze in time and start rendering the wireframes that make up his head and upper torso as well as the environment around him
Very true about search engines. Completely useless.
Love it that it gave you in the thumbnail a sojak face
I integrated Copilot into my coding workflow a little while ago, and it's been super interesting. On the one hand, it will provide autocompletions that are staggeringly complex, but on the other, half the time there's a little bug in them that you'll only notice if you read very closely. Also, a bunch of the time, the autocompletion won't even compile. I'm looking forward to writing unit tests and API's with it, though, as those tend to be paint-by-numbers type files.
I actually go days without doing a simple web search but when I have to ask a mundane, extremely generalized question it's impossible to get good results.
I as well use to search things on youtube because it was easier but now even youtube search is completely useless.
The only thing I currently think that Chat Bot AI is good for is getting technical information quickly that I could have figured out myself more slowly. That is itself a double edge sword because you don't get the practice, but it can take a 5 hour troubleshooting session and turn it into 1 hour. It's not perfect but it is the only real use I have found that isn't a novelty.
Wait people genuinely trust AI answer for facts?
@@b0leg23 no, but the fact is that the simple troubleshooting questions you ask are either correct or immediately obvious they are incorrect. So either way you save time.
processing and storage, you cant afford to train a model if you are not a corpo or a millionaire, two of my favourite entities XD
in the next decade there will be order of magnitude efficiency increases
whats next? yung luke smith rap?
It's not there right now but give it 20-30 years and AI will be so good it'll be impossible to tell what is made by man and what is made by machine.
This channel is a blessing
Unironically TH-cam shorts are now a good resource for small searches
I feel you one hundred percent brother! I also believe the real danger to any chance of preventing this from spiraling out of control and destroying the internet experience as we have known it, is that most people really don't care about it as much as they should.
Are there any actually good search engines in light of this information?
Wiby is cool simple search engine. I really enjoy the "surprise me" feature.
@@MrCrackerjack121 Wow. This feels like I've discovered the dark web.
My behavior on the internet is pretty damn eratic and random, but the problem is, that can be replicated as well.
Love the thumbnail photo. You should try to recreate it with an AI art generator
Based. I honestly was very impressed with chatGPT on day 1. The awe rapidly faded away. But I find that it is a very useful tool for software engineering. Takes away the boring stuff. You just have to remind yourself once in a while that it doesn't understand what you type in. 100% agree on the mediocrity of the modern web.
Can you give an example of it being useful for software engineering? Like a complete example please.
i'm reading the comments and now i think everyone is an ai, thanks luke
@@nickaharanas3932 I am HI
@@dankillinger Don't know about a complete example but generating controller code for openapi endpoints for example. We don't have very good tools for that in perl and it saved me a couple of hours.
In essence just spitting out boilerplate for stuff you don't find in the docs.
Manual intervention will always be required if that's what you're aiming at.
ChatGPT will outright lie to you, and will "apologize" if you catch it spouting pilpul nonsense.
I came from that search engine rant video, this is my 2nd video of yours looking for prescience lol