this takes are honestly dishonest. the point of AI is not comparing 1 promp or someone trying the tools doing professional work with it. there is already been massive advances in image generation keeping the character consistent between images and also control the posses with control net. i have seen many videos that look worlds apart from what you showed in this video, I wonder why you didn't use any of those? because it goes against the point you wanted to make maybe. is not gonna be normal everyday people playing around replacing artist. is gonna be artist aideed by this technologies who will do so. i work in data ing and we already see that the value is not to replace a worker but have 1 worker do twice the work in the same time. so you need less people. especially less juniors. for voice acting the tech is already there that is why they are so up in arms. since replacing voice actors can be done today, is already good enough for people to not be able to tell which is which.
i think you missed the additional tidbit that now ai is starting to inbreed itself to death, with previous ai outputs finding their way into datasets (and then in turn causing trained models to hallucinate)
You sir lack imagination/forecasting the future based on recent developments 😉 If you are unable to see the greater picture for the future, I'm sorry for ya. I'm glad that the ones being a little smarter know what the tech can do (and has been doing in the recent years) and are actually able to benefit from it instead of just undermining or ridiculing it.
i am 100% convinced Suchir Balaji was murdered. Corporations have been getting bolder and more blatant in recent years in taking out whistleblowers. we need to stand against it.
_"We hired private investigator and did second autopsy to throw light on cause of death. Private autopsy doesn’t confirm cause of death stated by police._ _Suchir’s apartment was ransacked , sign of struggle in the bathroom and looks like some one hit him in bathroom based on blood spots."_ Poornima Rao's Tweet from a couple days ago.
It's sometimes on a level of "stealing" someone's idea, which can sometimes be bad but for the most part doesn't matter much just like somebody making a battle royale shooter because they are popular. Because that's the only thing someone generating an AI image brings to the table, just some general idea, and if it's too specific then the AI can't even replicate it well, and nothing else in the process is a human creation unless you get into the more complicated AI image generation software which 99.9% of people don't do since at that point if you don't have any artistic skills you'll make the results worse. All "creations" are on a spectrum of how much human work went into creating it and for most AI images less than 1% of what matters was human-made in the entire pipeline
A 'recent' issue that I encounter often is when looking for any references to draw/model/... many of the results are AI generated, but the quality is high enough that I can't tell from looking at the thumbnail, so the workflow is significantly slower than before. AI is not only NOT making being creative easier, it is actively making it harder.
@@coondog7934 if you are talking about ai making being creative easier, that is fair. Its my experience (I always made all assets myself, so the advantage from ai is mostly getting non-vital code, not exactly a creative thing) (The other thing you could be refering to) Going through a feed and trying to find real references is slower no matter how good your perception, just like with ads in a feed. It takes up space and time
At least in Google you can use command “before:YYYY-MM-DD” to search pictures before specific and not specific date You should type it in the end For ai it’s “before:2019”
As an artist, back when I saw the audio models and the early image models, I accepted the idea that they would end up being very useful for my artistic input, that they would be like "my personal movie studio". Five years later, I have to write "before:2022" everytime I search for references with Google, and the closest thing to "my personal movie studio" is a 2004 videogame.
@@ali32bit42 Funny enough I do it with GTA San Andreas (it isn't as advanced as SFM, but it's fun to play with the limited engine). Though a very lazily done machinima does a far better job than AI, not gonna lie.
This. Even experts fail A LOT of the time. Most of what we are able to do is several speculations pointing out to too many different directions and then emotionally and mentally prepare for the worst "just in case".
9 днів тому+75
@mekingtiger9095 the more you progress in life, the more you discover that most experts are just experts at seeming like an expert
I forget the exact quote (and whoever originally posted it, sorry) but one of my favourite responses to this whole AI "art" trend was "Instead of doing the tedious tasks for creative people it's doing the creative tasks for tedious people."
When the mining jobs were destroyed in America, I remember when they were all told to learn to code. People accepted things like that as the natural progression of our society. Did they think it would stop for them? It didn't stop for us, no matter how hard we fought it.
Instead of helping the artists make more money, it helps non-artists _spend less_ on artists. This is very obviously the goal, and no one is shy about this. And while they are not in the wrong, artists can be really entitled bitches about this... I am a programmer. And while right now, these tools are helping me, I am not an idiot. I also see the writing on the wall. Juniors are hardly getting hired anymore, because of (speculation about) AI. Right now, we're totally fine. But eventually we won't. Just like artists won't, except a bit sooner. What is happening to me now is what has been going on with excellent tools like Photoshop, Blender, Unreal Engine etc. for decades now. Tools that are getting better and better, which let increasingly inexperienced people create "good enough" art. It should have been obvious for a while now that these tools will reach a point where they will do it by themselves. Anyone that was caught off guard by this simply wasn't paying attention.
@@hundvd_7 Except that ai can create code too. And the code it creates barely ever executes, because of course it can’t. If speculation about ai is what is what’s killing every industry, then you know what will kill ai? The fact that the end of the day, these companies all want to sell product. And software cannot buy product. It sure as hell can make “product” but all it does is flood the market. Bitcoin and NFTs died because they were incomprehensible to most, gen ai is comprehensible, but unexciting. They are trying to sell a niche product and acting like it isn’t niche.
One annoying thing about AI, is almost all the AI services charge money for tokens, that are quickly wasted by bad generations that totally ignore several key parts of your prompt. No matter how many brackets you wrap around them. You quickly start to realize that AI is not so intelligent. No chance of a token refund once it's been made...
Adobe took a slightly different approach, it charges for tokens but only after you’ve iterated and approved the end result. If you don’t like it you don’t pay for it. (Adobe Stock’s GenAI feature)
@@coondog7934 Correct. Also electricity is cheap. People have this idea that computers somehow are inefficient. I have heard some absurd takes that AI used the same amount of electricity as the airline industry or something and that is simply not true. They were citing the estimated electricity usage of all datacenters worldwide. So they basically were taking the figure of all the computing power of the internet, and private computer networks combined.
It will never have context, personal experience, or perspective. It can only regurgitate images. It has no meaning. It can’t create real art, only images. Artists are informed by culture and create culture. Ai is nothing but soulless images. It absolutely disgusts me how every company is forcing it on every damn thing. It makes me even more motivated to become a better person and do better in life in every way.
Same... and one day that will be valued, people will want a work they can hold, that truly is one of a kind (unless a print), something they can see the texture of the implements.
Even if the battle is won against AI the war is never ending. Even if corporations reel back due to backlash, they might try again in a more subtle way until they get what they want without anyone realizing what we lost.
What they want is industrialization of graphic design and other non-art things. Why would I pay someone 30 an hour to design icons for my website buttons when an AI could do it for 5 a month to do it in 10 minutes? And they will accomplish this. The Fiverr artists who’ve basically been overcharging for simple graphics will be proletarianized like everyone else.
Absolutely. I've seen a video somewhere, can't remember exactly but it made me realise how truly AWFUL the people in power are in their neverending seeking of money. It was about robots in movies. Have you noticed that sympathetic robots, ai and such have been pretty much always symbolic or an allegory to racism and similar? And the bad tended to be a warning. Now I can't remember a recent movie about evil robots, the most recent being the latest terminator (witch mind you, I also don't think is realistic, but I am talking about the perspective of the people making the movies), but there seems to be an uptake of good robots, without the racism allegories and such, in the ways that they are so much less specific and truthful, I guess. It makes sense, when you remember that Hollywood is trying to use this technology. Think Disney, since they own most of everything anyway, first they put a fake dead man in a not to prominent movie (Rogue One), and we laughed because it was ugly, but like the actor is dead and tHerE WaS nO OtheR wAY...we move on. Same franchise, more prominent series getting traction on the platform they were pushing for. We take it, the actor was still there, and it's better, tHey HaD tO MakE HiM yOuNgER SomEHow...we move on. Small steps they make. Scans of their actors they take, fooling us and them they won't use those without permission...without a contract...without pay...say if there is no contractual obligation and we have the assets, who's to stop us from not paying those actors millions "just" for their faces...ah, the public... So what are the evil conglomerates to do then to make it ok. Media is powerful propaganda and they know it. Kids jump in wells for teletubies, but how subtle can they be for the older people, who don't have much of an opinion on the matter, especially the young generations who don't know the grungy old sci fi and cyberpunk movies, but know a world of greater and greater acceptance. It is awful, and not all creators are in on this mad scheme, but they will support more creators that they can bend to their needs. It makes me sad, not for the fearmongering of a dark future, but for movies like wall-e, that could tell a story of how humans connect and survive through robot characters, while also having one of the best villains for that exact purpose. Unyielding and unmoving in it's code, not wanting anyone to love and feel and LIVE, knowing nothing else, but following a line. Because it will have to be marinated in the state of this weird war we find ourselves in. Robots were a fascinating storytelling tool, that now has to carry so many more tones and baggage because of all this.
@@gmodrules123456789 Why would I pay even a penny to a robot whose only works have no originality and are simply stolen? Why would I pay 30 to an artist? Of course, it’s because they’re human, and their works have originality, even if they - in some way - are inspired by other works. I don’t care how bad it is, I only care if the artist is human. People like you need to stop glorifying artificial -intelligence- stupidity because in the end, it’ll all fall. Art has existed for thousands of years, and your simple underperformance in them makes you seek that sweet sweet little machine to write in a prompt and expect something good. Commission an ACTUAL artist for that 30 you said, and then give the prompt you gave them to a machine, and compare. Which one is better? Of course it would be the artist because there is genuine emotion put in it. The “art” the machine pumps out has no emotion nor originality, and you only glorify it because you are too snobby and lethargic to even check an actual artist’s works, and commission them. “It’s too expensive” so is the AI?
The additional eyes on the back of the head of the cat unlocked a rare emotion in me that's akin to the Uncanny Valley. I'm pretty sure this is what it feels like to see something truly alien and frightening.
I can't tell you how many times I've seen AI-generated video spontaneously descend into a surrealistic nightmarescape. Its almost like there are malefic, otherworldly entities that are trying to come into our reality, and the prompts are akin to Goetic sigils...
I don't like body horror but I like to use AI to make pictures for fun so I can only make it short burst. So many weird realistic body horror images. If you look up gymnast or break dancer videos for Open AI and body physics will look really weird somewhat funny also really weird.
I miss back in 2022 Where ai was just a silly little thing that made stupid looking funny images Then a year later, it was everywhere And I wish it would just go away, it’s ruining and overtaking everything I enjoy
Me too, man. I remember that I quite enjoyed using websites Craiyon and that website that made AI generated pokémon who's name escapes me right now to make some funny, silly creations. It was fun for what I thought it was at the time. Safe to say my opinion on such websites have... _deeply soured,_ let's say, since then. I know better now. I wasn't making anything.
RIP Suchir Balaji, the OpenAI Whistleblower We come at the era of whistleblowers but FTC and SEC had to resort to mooch off CIA's drug money to fund witnesses protection while the tax money goes to Bitcoin, Microsoft, Lockheed Martin, Boeing, and Elon Musk.
It's hilarious how easily people have learned to spot AI. My parents are in their 80s and are computer illiterate, but even they will point at some image on a website and say 'Yuck, AI image'. 1 year ago nobody had this skill. As for your final conclusions, it's waaaaay too early to say where this will go, it's only been like 18 months. I don't believe the industry in anywhere even close to AGI, I think that's way more complex than they realise. But LLM generative AI will get a lot better, the real problem is that it uses so much resources they have to try and make it profitable right now. But it's currently worthless in all but a few niche applications. Even for stock footage it's a distraction. But in ten years or more... that's when we will declare this dead or alive. Right now is just waaay too early.
These kinds of technologies are always an arms race. Cinematic special effects that amazed audiences in the 80s look comedic today. So artists kept pushing the envelope until a new generation could suspend their disbelief. I suspect the same thing will be true AI. People will look back on these early efforts as cartoonishly bad, but the tools available in their time will seem like magic to us. Agreed about it being too early. The people writing it off as dead this early seem a little desperate.
agi is a ridiculous goal, and any people that think we're close seriously need to go touch grass, observe the ants on it, and think about how even tiny creatures are so complex
For real, companies are borderline openly murdering whistleblowers now. It seems like its all whistleblowers now turn up dead. Corps are hiding it at all, and it's a depressingly common knowledge fact that, yeah, they were probably assassinated
Us? No, that's the government's job because they made it harder in the first place. Espionage Act And letting the companies do whatever they want surely doesn't help that either
Seriously dude, this is probably why many people won't speak out about stuff they have inside info on. The risk of committing suicide with a bullet in the back of your head is just too high. Shits crazy, and it happens way too often to not realize there's an issue.
18:31 I've seen people TRY to actually use AI to generate pixel art, but the fact that the fundamental structure is flawed, just makes it so they can NEVER even get CLOSE to legitimate pixel art.
I think something ai slop has made me realize about real Art is that it is a form of communication. Every piece of art says something about its creator. Every painting, movie, show, book, comic, photo, piece of music, cave painting, napkin doodle, etc. even if its unintentional theres always something that is being said. We as a species have been creating art for as long as we have been talking to each other, I think that are abilities to communicate and create are the core part that separates us from other animals. Ai has none of that, it has nothing to say, it is fundamentally anti human.
AI art is the same. What it generates isn't a reflection on the AI, but on the prompter. AI isn't just there producing artworks. Someone had to tell it what to generate.
@@DeadFishFactory The issue is that depending on the prompt or the direct interference the user had on the output, the influence he had beyond just a basic conept is so minimal that "the message" might aswell be attributed to the AI itself since it takes far too many creative liberties separately from the user.
@@DeadFishFactory it's not, though. Two artists can have the same idea but wildly different ways to express it. AI "artists" will only be able to make the same generic slop
@@mekingtiger9095 If your prompt is generic, the AI will make it generic as well. The same generic prompt can produce wildly different results because it's starting from different seeds. There's no message and it's basically just statistical noise.
@@LineOfThy It's all in the prompt. Two artists with the same idea will likely describe it differently to the AI, producing different artworks. Obviously, if both are using the same LoRAs or even sharing prompts, they can arrive to similar AI outputs, unlike their artistic abilities which can vary wildly. Two people with the same idea will produce two different literary works as well. They won't exactly describe it the same way in a poem or writing, and that difference is all that is needed in the prompting for the AI to produce different works on the same idea.
I think I'm slightly less concerned about AI art than I am about AI writing. It's already becoming incredibly difficult to find information on niche topics without stumbling across generated articles with quite frequent factual errors, especially in the tech sector (which feels pretty ironic), as well as the capacity for AI to streamline scamming operations through TTS, which is getting increasingly convincing. All in all, it's just a ludicrous amount of resources dumped into something that i feel has yet to demonstrate a positive reason for it's existence and insane amount of investment - a while ago all this tech was conceptualized as a way to save us time and money, to enable us to enjoy doing more of the human stuff, and it hasn't really accomplished any of that yet - ESPECIALLY not for anyone of lower or middle class.
I think the only genuinely harmful thing AI has done is pollute search engines. My parents and even my somewhat tech savvy friends frequently quote Gemini AI from google because they don't realize it's an AI result. Literally a week ago a family member who is in his early 30s thought Kathy Bates was dead because he googled it and the first result was "Kathy Bates died in 2022" by Gemini.
7:12 ...huh. I mean, it's obvious that AI does scrape copyrighted art, but wow, that's pretty brazen. Not to mention that tech companies are only going to gain more power over the course of a few years. Yeah this is not good.
@@LineOfThy Why is it not theirs? If I carefully study the works of the art nouveau movement to the point where I can paint new ones, are my paintings my own? If they are, then why is it different if a computer does it?
Maybe the downfall of AI was really just the downfall of NFTs 2.0 all along. Just far more agonisingly drawn out with all the cryptobros who jumped ship from NFTs to it still being neck deep in the denial phase. The bigger they are, the harder they'll fall.
@@coondog7934 I COULD see major attempts at AI falling apart but that's a little ways off. Like, imagine if OpenAI had to shutter.. fat chance, I know, but if anything close to that happened, AI interest would probably die down significantly
I had CGWallpapers account and was astounded at the amount of tiny details artists put in their creations like tiny bits of screws and bolts, texts etc. and when I think about the amount of time and dedication they've put through, I become kind of sad that AI has devaluated their work for normal people who don't know the details and differences and perfection of these artists who have spent years to master 3D softwares and say: I can do it too now. the truth is, no, you can't. AI is not perfect like human art and would never be. I'm still always looking for human art and real photographs and never care about AI art seriously
@aolson1111 thats how most anti-AI screeds read to me. Also the interlude with an artist in CS Lewis' The Great Divorce where he flips out at the revelation that in Heaven, being a great artist doesn't get you attention or accolades for yourself because the the actual purpose of subcreation is glorifying the Ultimate Creator and never yourself. AI art is trash, but it just unmasks the trash that most contemporary art has become.
@aolson1111 there's something special about knowing some every little part of what you are seeing was simmering a person thought about and intentionally did. Like when you play a video game and go somewhere you think literally nobody would go and the developers hide something funny there... it just wouldn't feel the same if that was just created by a computer and not a human interaction like "oh that's so cool that they thought of that"
@@aolson1111he’s clearly referring to the soul of good human art, that every detail you see both has a reason to exit in the world depicted and looks good. Neither ai can ever truly achieve
As someone that works in 3D modeling/animation and even went to SIGGRAPH last year to look at the developments, even there I think AI is being overhyped. They had lots of fancy stage demos that made it look really easy and useful for environment building. But when I went to the floor show stuff and tried messing with it myself, the results were far less impressive. The output models are a topographical mess, similar to a photogrammetry model, which is basically what they are when you get down to it. They're generating 2D representations of the object then creating assumed depth from those images and making a model from that. The result is a mesh that looks kinda alright from just the outside but it's polygons are attrocious and the UV/texture is a wonky mess that makes no sense. And even if you make something more complex (i.e. a monkey on a bicycle) it's gonna come out as one hole piece just in that pose making it's actual use without major overhauling super limited. Whether that overhauling is actually less work than just doing it right from the start is a case-by-case issue, but personally I'd rather just make it myself and build it with the intended use case in mind. Also, the thing has some major limits in generation that I don't see them solving anytime soon. Biggest one being transparency. Tried to make "A brain in a glass jar" multiple times and it had no way to figure it out. Usually put the brain on or sticking out of the jar cause it can't make something transparent with another object inside it since it's building things out as a single blob mesh.
Dude, you are criticising tech that has basically been invented just a few years ago. Just give it a few more years time and you will not use anything else by then. BTW regarding the transparency thing: Basically 99 percent of all models are trained with .jpg images and those can by design not hold transparency information. So no wonder you won't get transparency right, you need to use a model that was trained with .png images instead. It is not the techs fault if the users don't know how to use it.
@coondog7934 Like how in a few years flying cars would be the norm? Or self driving cars? Or Crypto Currency? NFT's? Tech bros have a horrible track record with this
Actually no it wasn’t “just invented a few years ago”. That is cope. LLMs have been in development for decades. They just weren’t referred to as “AI” until a few years ago.
@@coondog7934 Transparency in 3D means by actual transparent material generation where the 3D object inside another 3D object can be seen thru. the 3D Model itself must have a transparent glass or similar material on it where the transparency is showing another 3d model inside. Imagine you tell AI to generate a 3D model from a 2D picture of a Brain floating in a Glass Jar, it doesn't matter the image format is png or jpg, its just an artwork. What the AI will do is make the 3D mesh of the Jar but once it will try to make the mesh of the brain inside the 3D jar, it will fail cause its just making it all from a single mesh. And AI doesn't even understand how Transparency even works in 3D (it works like it does in real life to an extent)
Ai might be a generational nothing burger. All the hype, all the corporations pushing it in my face, all the fears and it all lead to no significant change in my life or how I engage in art. I still watch shows and videos made by humans, I still play video games made by humans, I ignore anything that is made by ai and I talk with ai chat bots just for momentary fun and not as a true companion. It’s all just a nothing burger to me. Ai has never and will never replace my appreciation for real art made by real people to me. Edit: the nvidia stuff is pretty cool though
Same. AI is just best for the likes of shitposting and that's about it whether it's having some art generator make something born from a night terror or listening to someone's video of a cartoon character saying bad words that wouldn't be seen on a kids show.
Fr dude. My back catalogue of games, books and movies extend far back before AI was a thing. Im literally on a binge of a bunch of PS1 games rn. I'll be dead and gone before I even get around to the AI slop games that may or may not permeate the industry in a few years
@@redblue5140 Not sure that's a good idea, AI is notorious for making shit up because it's largely just guessing what the most likely next word or two is and doesn't actually understand what it's saying/doing bcus it's... y'know, just a program. Or what do you mean by 'act as google'? Sorry if I'm misinterpreting.
You nailed it. Generally speaking, AI generated works are created by finding other works related to the prompt and averaging them together. That's why the output always looks generic. You can only generate original-looking output by coercing the AI to average together works that have never been averaged before. But even then, the lack of control makes it difficult to generate anything really useful. And the more control you give the user, the more work they have to put in, the less they save from using AI.
That is the thing though. for people who use the tools to make things practical which require higher levels of control, it doesn't necessarily mean that they are looking to be faster than say a skilled digital illustrator would be to create the same thing. They are attempting to use wholly different skills to achieve the same result.
@@DarthT15then how do you find Inspiration? because i think artists arent so different in creating something unique, they look at things and then create something with that inspiration but can that even be a one time thing? or are we just copying one another without realizing
@@Crit-Multiplier Ask any author where they get their inspiration from and you'll get answers like "I was driving in a misty night not able to see a car length ahead." Ideas are cheap, they can come from anything. No idea is so insanely clever that it's a final product in and of itself. Execution is what actually matters. It's the hundreds of hours that go into writing a novel. The decade of practice that happened before working hours on a painting. AI cannot copy that. It can only ever copy an end product. AI cannot copy the muscle memory a painter has crafted that make their strokes unique. It cannot copy all the mistakes and failures that went into crafting a recipe for a cookbook. It can just guess at output that resembles the final product of all that work. So when you claim that AI is working like humans, no it isn't nor can it ever.
I watched a church full of boomers get emotional over an absolutely God awful AI generated Christmas short film on Christmas Eve. Uncanny movements, inconsistent style, six fingers, missing feet and legs, etc and the boomers were lapping it up. The boomer will keep this shit afloat.
I remember once someone said something along the lines of “ai art allows art to be more accessible to people” and thinking, no. art shouldn’t be accessible to you if your not willing to put in the time or at least the effort to make something.
That's exactly the same bootstrap mentality that conservative boomers have about everything else, though. "Don't work, don't eat" kinda thing. People have the human right to be creative. Expression is a fundamental part of being a human. Whether something is art or not has nothing to do with effort, and effort shouldn't be the metric by which expression is judged.
@@_B_E eh not really, I’m not saying no one should be allowed to make any art unless they put in x ammount of hours and x ammount of effort. I’m not even really upset about using shortcuts. But AI “art” goes beyond just a shortcut it takes out any actual expression or creativity a person might hold. When you’re just going to type a couple words and make a machine spit something out and claim it’s art, instead of trying to make something yourself regardless of one’s talent. That’s when I have a problem.
I'm gonna play Devil's Advocate here, though: Yes, art is and has always been theoretically accessible for everyone willing to grab a pencil and a paper and learn. But you want me to be honest? It depends a lot on *HOW* you train yourself. Basic trial and error is the most inefficient way to learn, taking YEARS to show any progress. Fundamentals aren't often taken seriously or have their importance emphasized enough, and 95% of all artists and guides I've found on the internet are TERRIBLE teachers. I've only started to see an improvement with SOME artists providing better educational material since this whole AI thing blew up and even then it is not widespread. Maybe if that wasn't the case, we wouldn't have so many desperate souls resorting to AI as a crutch for believing they don't have the "jeans" or skmething. This one's on artists.
@@Scrimblytimblo "art shouldn’t be accessible to you if your not willing to put in the time or at least the effort to make something." this is LITERALLY what you said. You're LITERALLY saying "no one should be allowed to make any art unless they put in x ammount of hours and x ammount of effort.". You're basically just acting like you didn't say the comment i replied to and then moving the goalposts lmao
Every possible way to use AI "art" (it is exactly what art isnt) to do something 1: make a shitpost (stupid thing) 2: pretend to be an artist (bad thing to do) 3: make footage of someone commiting a crime (bad thing to do) 4: create ads without paying an artist (horrible thing to do) 5: create spam bots (horrible thing to do) 6: make content farms without even making anything (horrible thing to do) 7: something that can help y ou learn (difficult thing to do) 8: cheat tests (thats only hurting yourself) 9: misinformation (horrible thing to do) AI can be mostly used for doing bad things and few good things, so in other words, AI is just Riskware
I don't think AI art will ever truly replace the creative process, because the only people who want to replace artists with AI are greedy, lazy techbros who have zero passion for art beyond its use as a commercial product, and thus everything they make is this gaudy, unpolished garbage that only resembles something man-made if you glance at it for less than two seconds while scrolling thru social media, just another piece of candy to mindlessly cram down your cake-hole. They don't care if their product is functional or beautiful or presentable, or even if it physically exists; they only care about making as much money as possible with as little cost to them as possible, to just press a button and have a computer spin straw into gold for them.
it cannot replace the creative process because theres nothing inherently creative about it. its essentially like using google to search for something but it generates what youre looking for instead. imagine how retarded it would be if we called people who used google search artists lmao
@tyujg7495. I have sort of the same experience, but with Photoshop. I sometimes use the Content-Aware fill tool to remove stuff from pictures, and it usually does a decent job, but I always have to touch it up myself to make it look more presentable.
Tech is built by people. Shock and horror, people have biases Humor aside, look no further than facial recognition. Years ago, the less Caucasian you looked, higher chance to not even be picked up as a person
Yeah, that was the first thing I noticed. It's like the meme of computer upscaling turning pixelated Obama into a white man (and that was 5 years ago...). You hated to see it then; you still hate to see it now.
@@RadikAlice Yeah, well known. When all the learning data you put into AI is pictures of white people, that's what it considers to be standard. It's amazing how 30 years ago, the first book about computers I ever got had "GARBAGE IN, GARBAGE OUT" as its first programming warning on page two, caps and bold and everything. Decades later, we're still shovelling garbage into the system and getting garbage out, especially as AIs reiterate on one another.
It's tech bros not understanding the fundamentals of what makes art ART. AI "art" isn't art; art is inherently human. An AI vomiting out what it thinks the prompts mean will never compare to even amateur artists. Part of the appeal of art, at least to me, is the imperfections in the line art, the colors chosen by the artist, and the appreciation of the time invested to learn a skill. The only people I think who will still use AI art in the future are going to be corpos who just need it "good enough" (looking at you Coca-cola) to post slop so they don't have to pay real artists what they are truly worth. Edit: Damn the tech bros are malding.
There most definitely is a spectrum when it comes to how much any kind of vision was being expressed when using AI tools. Just like with photography, cameras can be used to create art, but that doesn't mean every selfie and jpg in your phone is. Art part of generative AI comes from the human learning the insides and outs of their specific model so the "AI vomit" better matches it. I would compare generative AI to sculpting in many ways. You start with a very rough, general shape and then you chip away at imperfections. When you have exhausted the capabilities of one model, continue the process with a different AI model specialized in different things. There are plenty of human decisions being made when using generative tools if the user has the will and the capability for it. Photographers do something very similar too where they take tons of pictures, choose the one they like the best and then further enhance it with photoshop etc. This process is nearly 1:1 what one does with generative AI as well. Especially as photoshop itself has changed a ton from what it was 20 years ago with so many automated features. We are also approaching very much the point where actually well made AI pieces are indistinguishable from digital art to a layman, as the picture has been gone through with 1000% zoom and every weird pixel touched on with some other tool until the human in the loop liked the outcome. Because of this when AI tools get better it also skews the perceptions regarding what "AI art" is supposed to even look like, as the people who along the years get really good at using AI tools slowly fade away from public consciousness as "AI artists" due to the quality of their outputs, while simultaneously the skill floor is lowered as the programs themselves become less technical and more "plug and play", which leads to people who care even less about self expression using the tools. There's metric tons of nuance with AI art, and the spectrum of people using these tools is just as wide. At the more higher-end I'd say AI art is no less art than photography is. That too could be argued in similar vein to be "finding a cool thing and pressing a button, then claiming you made something kekw." But when you add all the technical aspects to it and take someone who is genuinely passionate about it, it becomes art. Even if your uncle's blurry picture of his car he put no thought into and posted on facebook might not be.
AI can generate something based on algorithms, it doesn't understand what it's doing, let alone feel it. When humans do art, they have an understanding of it and emotions attached to it - when you're happy you create different art than sad even if it'll be the same "prompt" to draw.
@@MidWitPride if you're gonna go down and tweak every single pixel, you might as well draw it yourself(especially since usually the output is ugly anyways)
@@player_8008 Art is self expression via a medium, and if someone likes to do it through one method and not some other, it doesn't contribute to the question whether something is art or not. Art is not defined by the effort it takes. If someone likes the process of generative AI more than doing more traditional digital art, then that's what their self expression is about. One could have endless "Why do X type of art instead of Y?" comparisons regarding methods.
@@o_o825 If we calculate a year, that value would be around 255$ million a year for OpenAI just to keep running its LLM. Meta spends around 50$ million running its servers. And it earned around 131$ billion in ads just in 2023. OpenAI revenue from its users was around 300$ million in the month of August 2024 (Source New York Times). Which would push it around 2.1$ million a year in average. If anyone don't trust me, search for the values yourself please. The 700$K/day I got from the channel Cold Fusion
I’m an illustrator and for the first year I was terrified, angered and flustered by ai art, but then I realized something. Creatives develop tastes. A palette of media that influences their creative output. The reason why the majority of AI art output is so banal is because the people abusing it have not developed any taste and interact with media in a rushed encyclopedic way that is counter to making anything of creative value or merit.
The bottom line is that "AI" (LLMs) *can't think. They don't think.* An algorithm, however complicated, can't mimic the intricacy, creativity or capriciousness of actual, genuine thought. It was always the next bubble for tech start-up land to dupe investors after "NFTs" crashed, in 3 years I guarantee you will see the amount of AI crap being pushed greatly diminish as companies realize there wasn't actually that much money to be saved or made. The AI stuff that will continue will suddenly become less prominent and more in the background, dedicated to very *very* specialized and simple tasks, which is what they're actually good for.
So what exactly is stopping a multi-billion parameter neural network from thinking? Do you think the human brain is magic and the process of thinking can't be replicated?
@@alainx277there is no reason to assume the current method we are pursuing is how the brain works. It is entirely possible we have made something that doesn't replicate the brain at all.
@@alainx277it is you who believe in magic. LLMs lack the qualification to be considered life, which is the basis for conscious experience. "Emergence" does not com just from complexity, but from specificity. No one can claim that sentience comes from anything but biology and be scientifically accurate. To think that your computer software is alive is believing in "magic".
@@alainx277 The problem is that neither you nor me know how thought actually happens. The human brain is opaque, in a number of ways. How do you replicate a black box? Sure, you could make the stupid thing more and more complex, add a billion or a trillion parameters, but that doesn't guarantee you're any closer to replicating thought. What you and a few thousand other people are proposing is the equivalent of firing an increasing amount of shotguns into a dark room to hit a target that may not even be there.
@@alainx277 neural networks are just matrices we’ve tricked into predicting data using matrix multiplication and dot products saying it thinks is like saying a calculator can think because it can do tedious math problems or that a camera can draw because it can create a image of whats infront of it image generation programs cannot problem solve, critically think or be creative let alone to the extent as a human outside of image generation, programs that somewhat problem solve like atlas’s ai created by boston dynamics, is still very much limited in terms of problem solving and isn’t designed for complex problems
>2022: haha dall-e is so funny it can't draw very well brrrrr >2023: AI art is going to bring the end of society and take away all of our jobs. Hyperconsumerism is imminent, late capitalism has forced us to discard our most human qualities in order to buy more. We are worth little as people in this system without opportunities. If our ability to create is being taken away, what's next? >2024: Oh that's, that's another content farm that flooded my feed with AI Shrimp Jesus Slop. It's the fifth bot I've blocked in a day.
Bet if someone got hold on one of those mechenical arm robot thingys its prolly possible to do this but with craptons of programing and its probbably easier to do by hand
the instant i find out something is not made by humans it becomes 100% less impressive, the people around me and those closest to me feel the same. so i don't think it devalues human made art at all. and honestly that comforts me. we (for the most part) still desire seeing the display of human skill that it takes to make art and i think as long as we keep that opinion we will be ok for a while.
And what if you use AI together with your own skills to create something unique? Would that be 'less impressive' as well? Your opinion seems to be rather one sided. AI is a tool and in the right hands it can do wonders. Just like in the wrong hands it ends with stock footage and sloppy stuff.
@@coondog7934I mean yeah? People still value hand crafted anything over factory made. Both are made by humans, but one has a hell of a lot more effort put into it.
I like the idea that at the end of everything that a corporation has gone through in order to cut costs by cutting employees, they'll just have the CEO left as the only employee and he is sitting in a board meeting with the shareholders and they say that they need to cut costs further and do a round of layoffs, then they just look at the CEO and go like, "Thanks John for your service to our company, but we will need you to fire yourself"
I praise AI in one thing only... It actually encouraged me to pick up a pencil and draw. I always liked art but thanks to ai artists acting like the most annoying people i would not like to meet. I just picked the pencil to see if its that much of a struggle and noticed that its rather relaxing to draw even if its just random shapes to practice drawing bigger things.
AI is so useless. All the problems it could ever possibly claim to solve were already solved by things like templates for writing, calendars on your phone, alarms on your phone, online interest groups, a huge amount of tools in art programs, brush packs you can buy for less than a dollar or even get for free-- all those things already existed and have been used and praised by artists the world over. AI programs demands you pay hundreds of dollars for something that does all that significantly worse. Why would I want that? I've been saying this is the same as the NFT craze for years, because it is. We know exactly how this is going to go, we already know the lifecycle. It's new, come corpopainthuffer liked the sound of the pitch, it's uselessly pushed into things it doesn't make sense to push it into, the scammers get it, and death. Also the internet is full of the most degenerate, hardheaded, incompetent kinds of people humanity has to offer, and you're training something that cannot read or see or comprehend off their contributions. Ask the chatbots they call AI how many Rs are in something like grooseberry. 999 times out of 1000 it'll tell you there's only 2 because people who were having a brain fart have asked about the berry part so many times on the internet. I used to illustrate that with the word strawberry, but then people kept correcting it so now it sees the full sentence as it's own thing and pegs that specific instance of symbols to mean three. If that make sense. Either way, you can't even trust it to help you in the most watered-down, innocent instances. It's not going to help you with your grammar as an example. more people having trouble with grammar are talking about it specifically with specific kinds of questions than people who can offer flawless advise are offering specific kinds of answers with specific ways to be the most correct. So you got more people asking: "would I use a semicolon here?" than there actually is a solid, repeatable answers to that question. Because that answer changes based on the sentence the questioner is actually trying to write, but the question is the same regardless, you are more likely to get bad advise than good. The AI is going to pick an answer that might not apply to you because it has no idea what the rules actually are, it knows what the most common flags are plopped down after a specific series of flags. And art is the same kind of flag-flag-raised situation. Your generated imagine of Scooby Doo has his schlong out because the most common training response to low lighting and Scooby Doo come from terrible corners and it doesn't get that you wanted a film noir Scooby Doo, you are not interested in dogs lipsticking.
@@YourWaywardDestiny You seem like you've got a lot of personal things to sort through bud. It's clear that you're blaming AI as a scapegoat for whatever personal failings you seem to be avoiding. I suggest you find a healthier avenue for working through all that.
@@YourWaywardDestiny Prepare to get your brain fried with the capability jumps this year. I read your long winded piece of nothing comment and it's clearly you that has no idea what they're talking about. AI is already much smarter than you. The r thing is a tokenizer issue and not the dumb shit you're talking about. It's also an expert translation system that is flawless at grammar in many languages.
Haven't watched the whole video yet, but your fundamental error in your previous argument was that you were looking at art from perspective of profit, saving time and making money
@@mrjaman3752 "Efficiency, efficiency, efficiency, efficiency" Tech/AI Bros see art as a mere "market product" to be commodified and not as, well, art. Reminds me of Zoe Bee's video essay of "In Defense of Inefficiency". Really good one. There's more to this world than just minimizing input for the pure sake of it.
Mentioned techbros and cryptobros shart out a borderline useless product/service like juicero. Only then they try to justify their thing by making up a problem that their thing is supposed to solve. None of the ai/crypto/techbros would ever understand that artists enjoy the process of creating art and there is no problem to solve for the artists. Artis wouldn't ever want something that generates "art" for them.
What so many AI users don't get I that Artists have a deep love and appreciation for the artistic process, not just the end result. AI will never replace us.
@@joshwenn989 Mona Lisa is inferior to the most mediocre of AI art, AI can generate art as beautiful and distinct as the entire Sistine chapel in seconds, AI is already where so many of the greatest images ever made are and will soon be if it isn't already the majority of GREAT songs/movies/images.
Does it? I can think of plenty of art where the artist was just slapping on paint. There is some human art where fine care was put in down to the last detail but it's in the minority.
@@peterhoulihan9766 What he means is the only answer on why an AI does anything is because it's what it was programmed to. There is no thought process, no idea, no perspective that's being communicated. That's a big deal if you want to avoid being generic.
@@joshwenn989 just because we don't know it doesn't mean it's not there. there is a story to every piece of art to what a human artist puts in. perhaps the artist was frustrated, had bad brushes, really wanted some color in a certain place, or just didn't care and wanted to fill the space with something. it's a nice time to look and appreciate art and try to figure out why the human did what they did. part of the art is not just the result of what is creation, but how it was created, and there is potential for a lot of appreciation for the results from unorthodox techniques like vomiting to produce something that is, subjectively, pleasing to one's eye. even if it isn't pleasing, it can be impressive, and if you don't think so, i'd challenge you to do that technique and show us the result if it's so unimpressive. one person i'd like to point out that produces fairly pleasing art in a fantastic way is tomwellywells (i've primarily found him on tiktok). his creations utilize spray paint and objects that cover the canvas in interesting ways that come out to producing very interesting looking planets, and before seeing him, i had no idea one could make planetscapes like that. i am sure if someone knew the technical reasons why llms do what they do, and the data that an llm was trained on, perhaps there might be a similar experience with regards to figuring out why something was created in the way it was, or if they happen to be familiar with a lot of artists throughout the world, if they can find where certain artists' techniques broke through. yet, once i know something is generated by the llm, even if i thought it was created by a human at first, i immediately start to dislike it, even if the result is more pleasing than some humans' results of art, because to me, the story of the process that drove it to where i could possibly see it carries a weight that text-to-image generation or image-to-image generation simply cannot carry. in the same way that watching something live and watching a recording of something that was live carries a different weight, human-created art and llm-generated mimicry has an ineffable difference that is meaningful to me. with llm generated mimicry, all these meanings from the human-created art it was trained on blend together to the things the trainers of the model selected for among millions or even billions of pieces of art, devaluing the stories behind them while wasting untold amounts of power primarily driven by a motivation of profiting off of devaluation of those stories. you could argue that's the meaning behind it, to waste that power in pursuit of profit by producing a generic art-like result, and at this point it's then known and the point has been made. they should stop before they continue to contribute to further ecological decline where profits will no longer matter, but i know they won't unless it becomes unprofitable until then. honestly, with the advent of llm-generated mimicry where companies are trying to remove the humanity from an already underappreciated profession, it's given me a deeper appreciation of that humanity behind the art, and that is why i don't call llm-generated mimicry art. art requires some creative spark and stories behind every word, stroke, and splatter, and while i won't say it necessarily requires humanity (some fish, notably the male puffer fish, produce pieces that i envy to create myself; some people believe our world to be a divine creation), i will say it requires a creative spark and story or several stories that i primarily associate with the humanity behind art.
The AI's own success also poisoned it, at least in terms of art. So much AI art was produced from mid 2022 onward that it's since flooded the internet and an AI scrapping the internet is going to pick up ever more AI art
Such an amazing society we live in! People work all week, overtime, do the same boring repetitive tasks while AI, a non living being, is being creative, making art, having fun! Truly, peak of human society!
Every time AI "art" defenders come out of the woodwork trying to analogize about how it's similar to the human process, they expose their cluelessness about the technology they glaze daily. Those arguments could be meaningful when talking about a hypothetical artificial general intelligence that has long term memory and an ability for abstract thinking which is necessary for imagination and composition (a lack of this is what you describe at 10:55). I don't agree with John Searle (of Chinese Room Argument fame) that this kind of broad artificial intelligence is impossible, but LLMs are not it nor will they ever lead to it. I think LLMs are a dead end in the progress towards such a technology.
Altman just wrote they know how to build it(AGI) today, believe him or not but OpenAI delivered great systems(except maybe Sora but that's hard to deliver to millions of users at scale so this is just the scaled down turbo version) that are already very useful for many domains. Expect very strong AI systems in Q3 this year and ASI in 2027/28. Things are gonna get wild fast. Remember to never use never in this domain. You will be proven wrong very fast. Send o1 to 2015 and people would say it's AGI for sure. Moving goalposts is a fun game until there are not any left to move.
@@therealOXOC Altman has been saying variations of that since the very start. It's not backed up by AI researchers from third party organizations or academia, so I have no reason to think it's anything but investor bait. I have no idea why you say I'm moving any goal posts either, as if you knew me. I have been saying what I wrote above for years and I don't see any serious sign that I will be suddenly proven wrong in the near future. Or if you mean that by some mysterious standard o1 would count as AGI a few years ago, that's nonsense. It's not about relative performance, o1 is simply not a general intelligence. Also artificial super-intelligence within 3 years is straight up delusional.
@@therealOXOC Altman specifically said that they now will consider any model that can make 10 billion dollars of profit AGI, regardless of its output/inner working. If anything, _he's_ the one moving goal posts here.
Nice. I was one of those people who brought these types of problems up with you on the original video, so I'm glad to see the general consensus/doomerism on AI's "inevitable" take-over shift over time. The more publically accessible a lot of these tools become, the more duped the public realizes they've been over the past years, it seems--and the more tired of it they are in general.
As someone who makes androids, AI is vital to my work and acheiving my childhood dream. As someone who loves art and likes to see work from talented people I admire, I have a burning hatred for it. May the first victims of the robot apocalypse be those who exploited and abused it
@@l1ghtd3m0n3well assuming AI can read the whole internet in less than a second by the time we make it sentient, maybe it can tell the difference between exploiters and real human beings (Maybe finally all of my info probably being shared around the internet by big corpos will finally pay off)
How has AI art prevented even a single talented person from producing artwork? Every other new artistic tool (photography, screenprinting, CGI etc.) has only made the creative industries stronger and more valuable. I seriously doubt this is going to be different.
I’ll admit, I like robotics and animatronic things. But I’m against AI art (cuz at least there were people involved in making those animatronics!) I realize this may not be the best argument but I feel double standardish like that sometimes. 😆
Man, you know what I wish for? If AI means that one person can do the job of 10, it should mean we get ten times as much cool stuff (or get it ten times faster), rather than 1 employed person and 9 unemployed people.
This issue has existed since cotton weaving machines , why pay more people when you can just overwork one because they can technically do it by themselves now
That's kinda how it works. The endless march of industrial progress means that your wage gets you access to so much more cool stuff in 2024 than it did in 1994, let alone 1924.
@@thelight3112 Eh, and yet most people in the United States have to work a lot just to live from paycheck to paycheck. I'm sorry, but when most people's financial lives are this unstable and are so close to being kicked out their homes, suddenly their microwaves don't seem so appealing. Technology alone doesn't mean much if there isn't *social* progress aswell.
one of the biggest issues with artists I've talked to is people being unable to put what they want to commission into words. entering a text prompt is just the same problem with faster, worse results so most people will probably never get what they want from AI generators. And imo AI generative fill is also useful for quick, small changes with images and there's def more tools that are actually useful. But yeah, it's not impressive to generate 5 seconds inconsistent of video while consuming the power equivalent of a nuclear power plant anymore.
Graphic designer here, the thing you said about AI images in the realm of stock video is spot on. I work with stock image libraries every day and it’s wild how many AI images there are on platforms like Adobe Stock (thank GOD you can opt to exclude them from search results)
As someone's who has been feeling a bit of existential panic about AI in the past couple of weeks, this really helped ease those concerns, at least temporarily. Thank you
20:08 this is exactly it: there's a use case for this type of tech, as a tool for creative people. But the people making the tech - and more importantly _funding_ the tech - are only concerned with cutting out creative people from their operations. They fundamentally don't understand what makes a competent, talented Creative Professional worth keeping on the payroll, and will happily kneecap their entire company just to lay them off.
There’s also the fact that since the whole internet is fully flooded with sewage, there’s nothing left to train the models off of. The artists have nothing left for AI to steal from.
There is a fair probability that artificial intelligence will attain the capacity to utilize its own output for training without any adverse effects to it.
14:55 the funny thing about this is that it is only low compute. On high compute it costs thousands of dollars per task, which means that OpenAI spent over a million dollars on the Arc test AND it was tuned for it. Looking at the examples of what o3 failed on the Arc creators blog makes me question the future capabilities of this technology.
I got to say the Advent of AI generation has made me much more appreciative of amateur artists and so-called Outsider artists who do not have a technical competency because they're the only people who can make something like that
The same way people adore those who drive around in horse carriages instead of cars because they can and know how to although they don't really benefit from it. Thats how it is supposed to be.
@@zombieoverlord5173 If you disagree you need to provide arguments otherwise you disqualify yourself from any discussion. Thanks. Until then my example totally holds up.
@@coondog7934 A.I art used to generate images is not making our lives easier and the arguments against it are not because it's new. Art made by A.I will always be bad and it's helping no one.
@@zombieoverlord5173 Au contraire, mon frère. My life (and hundreds of thousands others) already got better with it. It helps me a lot with my work and grants me time to focus on other important things on the project. Quod erad demonstrandum. I told you my example would hold up.
It gives people hope, I guess lol. This is the worst AI will ever be, and all of the kinks could be solved as AI gains the ability to improve itself and teach itself with artificially-generated novel data.
@@o_o825Not necessarily. As AI trains itself on AI generated content, it will cannibalize itself. Also, what does your reply have to do with the comment?
@@o_o825 Problem is that AI generated crap is always a little worse than the original, if AI can generate images that are 90% as good as the training it and them is trained on that it made it is only 33% as good as the original in 10 generations.
Say what you will about AI content only being a third as good… No one’s talking about how AI emits far less greenhouse gases on generating content than a human does for the same work. ;)
ai art is like a pleasure cube. It can be just as aesthetically pleasing as a human made artwork, but it will never have the same impact. Real art is meaningful for more than just its looks, the fact that someone thought it out and created it is significant. AI art is like if modern art was actually as simple and meaningless as detractors think it is.
Said by a someone who plays Fortnite, in comparison to which even the most brainless AI generated coco melon doodoo videogame would be better. A bunch of children who are responsible for making such a low effort game so popular that it ruined the whole industry flooding it with low effort garbage games and micro transactions. Ironic.
@Buzz.co2 And BTW, it’s all fine with me that you play Fortnite. That guy says that Fortnite “ruined” the gaming industry and that AI slop would be better… wish granted! AI tools advance to the point where entire video games can be “created” with a prompt, turning the industry into an endless morass of shovelware the likes of which the world has never seen. And I too pray that this never happens. All these GPUs and gigabytes of RAM, enslaved to crap out another fever dream nightmare. I’ve seen computer programs which are less than half the size of this comment, yet have more infinitely more artistic worth than the next Scarlett Johanssen cabin crew shrimp jesus
It may not completely ruin the professional side, but enjoying art as a hobby (as both creator and appreciator) has never been worse. Having art made by real people hidden underneath gallons and gallons of AI slop is discouraging for both sides (especially when the AI slowly gets good enough to make it more difficult to tell the 2 apart). If this problem only gets worse, perhaps future generations will see art in general in a much less appreciative light, who knows
I mean corporations don't really care about how generic the product of AI is, I realized that I actually feared a world where people don't really care if something is SOULLESS slop
Knowing how tech bros are it kind of is inevitable that they use computers to compensate for their lack of creativity and humanity. Not saying it's a good thing, but "AI" "art" was inevitable the second we discovered how to synthesize data...
AI is not gonna replace artists. People who genuinely value quality in art can see how sloppy AI is. Companies who will change to AI to save money weren´t very interested in treating artists fairly to start with (which might not even happen with the new price hikes) AI will most likely become a tool to speed up projects, while not fully replacing humans out of the equation.
Looking at how generative AI works (training on existing data to try to imitate it,) you will quickly realize that while AI can generate things, it can't create something genuinely new, it can only attempt to emulate us. You still need humans to make things that have never been seen before. Relying on generative AI to create stuff is essentially pausing the progress of innovation. Also, there is the whole problem of weather generative AI is stealing attribution from creators (which I believe it is.) Additionally, it is very hard to believe all the hype about AI, as things are very over hyped, and the progress of our current methods of doing generative AI seams to be slowing down, and the hope of being able to rely on training AI with it's own output has proved to only lead to the AI being worse. The funny thing about this is that generative AI is often trained on just taking a bunch of content found on the internet, and training it on that, but because generative AI has been so hyped up, there is a lot of content made by AI on the internet, which makes it hard to find training data that doesn't contain content generated by AI. Essentially, AI will be stuck in the past, it is potentially stealing from creators, it isn't as good as people make it out to be, and it's preventing itself from progressing.
9:45, thought about this some time back, origional art is ispired from life itself. Imagine a made up creature from fiction, it's usually a mashup of other cratures/elements that already exist irl. It's like the whole 'imagine a colour that doesn't exist' thing. Interesting stuff. I think tripping on shrooms might be a way to create something inspired by nothing.
You make some good points, particularly the way you articulate the issues of control and continuity. But I’m consistently baffled how people mistake current challenges for fundamental limitations. Speak to me of fundamental limitations after 50 years of stagnation and I might believe you’re onto something. After 0 years of stagnation, it’s baseless to claim that.
This key issue as to why the AIs can't produce content you are showcasing is that it's trained to be generic in order to appeal to the greatest number of consumers. That is why when an artist starts to input more content into the prompt the level of intent is never fully matched. There is skill in learning models for their nuances and how to talk to them but it is always specific to the model. I started leveraging different AI image gens to help me speed up my workflow as I found it gave me the ability to spend the same 10 hours working on fine details of my characters rather than spending 2 hours on details and 8 hours on background plates and poses. As a 15+ year digital artist and graphic designer I welcomed it. BUT, compared to the friends I've made who didn't come from my background, I'm excited to see them exploring their creativity where they previous were unskilled to do so but where I may generate 10 renders to find a suitable source to alter from they will generate 100s and still never feel good about it. In painting helps but so much of business and commissions come down to requirements and intent and AI without post-processing will never replace professionals. However anyone (other than stock photographers because they did get screwed) who "say never will I ever" is manifesting their career failure. I believe a good movie can be made with an AI generator but to get there the engineers CAN'T use a general-purpose model and first train a model to build the entire world within it. The characters, the art style, the POIs, the details within POIs. It must have spatial and temporal memory. At that point, you can feed the model a script and let the world play out. It would be a huge undertaking and unfortunately if successful mean a larger explosion of sequels as most of the following movie costs were already covered.
The idea of no one playing instruments, writing stories, painting etc. sounds like hell. Art exists for arts sake, we make art for the love of the game. Fuck generative AI so hard
the terrifying truth is that the generations that grow up in a world where AI is normal likely won't be able to tell the difference, or at best will not care, because they're used to it existing.
People who say A.I. art is supposed to replace art fundamentally do not understand what art is. Art is a form of human expression, always has been. Art is used to tell a story between humans. A machine is not going to tell a human story to humans. A.I. will try and find an assimilation of that, but it will never understand how to convey that. Art has to be crafted, art has to communicate. Art is made with a deliberate purpose in mind. In the end, art is made to tell a story(be it something personal, or even just an advertisement). A machine can't understand that, so it just makes copies.
Guess what, noone really gives... The end result is what matters and if that looks at least close to human art then thats all we need. It is going to look a lot better in the next few years.
@@coondog7934 All we need? How about want? How about desire? How about actual taste and style? It does matter. We're not all eating vanilla ice cream, my dude. We have discerning tastes. No machine is going to figure out taste or styles or wants, and neither are the techbros programming them.
@@rawjawbone You are absolutely right but still, who cares? Maybe you do, but the masses certainly don't. And noone takes painting away from you, no matter the technology. So you can still follow that passion as a hobby. If you want to look at human made art you will always be able to do so, since there will always be hobby artists left. For the big majority (and eveyrthing monetary related) AI generation will do for the most part (YT Videos, Games, short movies and maybe even entire full length movies and so on in the future).
@@coondog7934great can’t wait to play a game with 0 soul made by a big company because they need to save money so the ceo can buy their 5021th mansion and not pay their workers a cent
You fundamentally do not understand incentives in the world and what normal people think about AI. Industry is driven by profit, not integrity. Your worldview is childish and annoying
It doesn't matter how good AI gets, that will never stop me from writing. I write because I like it. Having a machine generate literature would be pointless, because the whole point is that I enjoy making it myself.
Exactly. No one is forcing anyone else to use AI. I use it to research for my world building and to evaluate early ideas. Having someone that can criticize my messy ideas without them stealing them or laughing at me is invaluable. Add to that a free "expert" consultant that can save me hours of scientific research for my SciFi and I'm sold. Actual writing is on my hands. If I ever let AI write something it would be the dialog of one unique type of character: an AI XD
Of course they did but they needed to. Without that they could have never developed their AI since it needs a lot of training data (and buying all those data instead they could have never afforded). And if artists willingly put their work out there for free for everyone to see they shouldn't be confused someone using it.
@@coondog7934 if i make a work of art and show it to you, you cant just steal it and call it your own. copyright is a thing that exists and i feel is often overlooked in conversations about AI.
@@coondog7934 There is an obvious difference between displaying your art for the general public to see, and having that work used without your permission for a product.
What irritates me so much about AI above all else, is that it had potential to be an artistic tool. Stable Diffusion Deforum, DiscoDiffusion, VqGAN+CLIP, they have all been forgotten in favor of pseudo-photorealistic generic crap generators and shitpost machines. The videos made by Linkin Park, Architects, Die Antwoord, Periphery, etc. were masterpieces in using AI as an expressive tool. A Doctor Strange-style superhero film with AI-made psychedelic magic FX made with frame-by-frame img2img StableDiffusion or VqGAN+CLIP would have honestly revolutionised many artistic fronts without costing trillions of dollars and without displacing any jobs or generating any controversy. BUT NO! It had to become the new iteration of the laziness feedback loop of pointless nothing our culture is right now.
The main issue with such tools is patents, if someone invents a cool new way to use A.I. as a tool they will patent it, sit on the patent for 20 years, and only after 20 years we'll see the technology come to fruition. Kind of like how we could've had a 3D printer revolution in the 1990's, but the inventor refused to do anything with his parents and made extremely expensive and inefficient machines. Technology would unironically leap decades ahead if countries would adopt a "use it or lose it" mentality to patents, unfortunately, most international treaties explicitly forbid this.
Out of all the things we could automate, we tried to automate art. What next? Robots that watch movies or enjoy meals for us so we don't have to? The premise of AI art was nonsense in the first place.
because they cant say the real premise: "we do not want artists to be able to make a living because their skills to critically think are too dangerous in today's world"
As much as I was interested in AI a while back as well, the fact that I don't touch the stuff anymore is a sign that it won't be as helpful as I initially thought. I mean, I have used it to (partially) create the character in my profile picture. But since what I was looking for was a different shape for my head and stuff, I didn't need to use it any farther than this. Every other time I've used it, it was to generate stuff for self-interest, or to get an image that I want to see. Then again, I'd rather look for stuff other people made, just because they are actually refined/crude enough to feel something.
As someone looking to get into the game industry as a 3D artist, the future of AI potentially replacing the skill sets is pretty scary. In theory, AI can be a great assistant in cleaning models, or even handling retopology is attractive, but seeing AI just make a model worries me.
The issue with AI generation of 3D models is that it also kind of sucks. The mesh is all over the place, it is insanely unoptimized due to a stupidly large amount of polygons, etc. Like it seems useful but if you had to use it in a game you'd have to change so much of it you'd be better off making it from scratch.
And for how long do we generate 3D models? Exactly, at max for a few years now. You do realize that tech like that needs a lot more years to become a lot better before reaching perfection, right?
@coondog7934 Yeah and we were supposed to already have reached perfection with image and video generation, yet despite the advancements every single fucking AI image looks the exact same and has to get a billion corrections done by humans to fix. There's only so many times techbros can say "DUDE, just give the technology more years bro please you don't get it." before I get fucking tired.
"Why would someone spend tens of thousands of hours learning to do art" The way this is phrased makes it sound like the only reason to do art is for profit or social media likes
Exactly. AI is not taking art away from you or me. If someone wants to follow that passion, just go for it and stop blaming technology. It still is a really neat and fulfilling hobby.
He meant professionally. art is a dream job of mine. I want to do art for a living because i love it and ai is taking away my dream job and pushing me into jobs i'd fucking hate doing.
I think you are right about the control limitations of Generative AI not being technical. The Tech Bro view is that inevitably as the tech improves the control problem will be solved. But this view ignores a basic conceptual flaw in the entire enterprise which is simply this- people using generative AI don't WANT control- they want the AI to do the job for them. That is-after all- the entire point of using AI. So to aspire to be an 'AI Artist' is to be in self contradiction- no real Artist wants to hand control over their art to someone or something else- they want to do it themsleves- they WANT control. So when someone says they are or wish to be an 'AI Artist' what they mean is that they like the idea of being an 'Artist' but are not really interested in mastering the skills and knowledge required to actually become an artist. Certainly it's true to say that there are many 'solutions' available to the AI Artist to try to square this circle- we have 'image2image', or 'inpainting' or 'style transfer' ect- a growing array of tools and techniques that attempt to solve the self contradictory problem of how you both give up control to the AI while at the same time impose control on the AI. But this is all nonsense- if the AI Artist really wants control the solution is simple- purchase a tablet and stylus and learn how to draw. However this is not what they want either- this is TOO MUCH control! The fantasy seems to be that somewhere there is or will be a 'sweet spot'- a goldilocks zone wherein the level of control will be not too little and not to great- that the technology will evolve to the point where 'something magical happens' and this technolgy will somehow do the impossible by granting granular control over a self guided autonomus machine. It's this logical absurdity that lies at the root of generative AI. We want machines that offer us total control without the need for us to exercise that control. The perfect AI is the one that does exactly what we want without being guided by us to do it- but how then does it know what we want? Language-it seems- is the answer. People used to beleive that certain combinations of words had the power to change reality- we called them 'magic spells'. Most people no longer beleive in magic- but we now have a new kind of magic spell- it's called 'The Prompt'- and in place of wizards we have 'prompt engineers' whose arcane knowledge of AI allows them to call forth from these machines all manner of wonders on demand- at least that's the story we are being sold. But there's a problem. If words could really describe Art in a complete and granular way then reading a description of a picture would be the same experience as actually seeing it- but this is not the case. The reason a 'picture is worth a thousand words' is because words cannot really express visual experience. For the same reason you can't drink the word 'water' you can't really 'see' a picture just by reading a written description of it- you have to actually see it. But if this is true then how can we tell our AI art generator what we want it to make? If words can't really express pictures how can our prompt engineer communicate to the AI the exact picture he wants it to make by using words? He can't. The 'control problem' of generatve AI is a linguistic illusion rooted in the idea that words really can define reality- and in any case a real solution already exists. There was no shortage of skilled Artists before 'AI Art' came along- nor was there a shortage of musicians, or writers or any other creative professional- people like doing this work and many are very good at doing it. Generative AI is fake solution in search of a real problem to solve- which is why Meta are going to turn facebook into a wasteland of bots talking to bots talking to bots- not because this is a good idea- even for them it's a very bad idea- but because they are desperate to somehow prove that generative AI has some kind of actual value, that it's more than just a vast macguffin being used to extract money from investors to keep the bubble growing.
Go to ground.news/husk to get 40% off unlimited access to the Ground News Vantage plan.
AI is just the next ".com" bubble. There is value in that, but not now. We will see it crash and then slowly develop properly. Great video!
this takes are honestly dishonest. the point of AI is not comparing 1 promp or someone trying the tools doing professional work with it.
there is already been massive advances in image generation keeping the character consistent between images and also control the posses with control net. i have seen many videos that look worlds apart from what you showed in this video, I wonder why you didn't use any of those? because it goes against the point you wanted to make maybe.
is not gonna be normal everyday people playing around replacing artist. is gonna be artist aideed by this technologies who will do so. i work in data ing and we already see that the value is not to replace a worker but have 1 worker do twice the work in the same time. so you need less people. especially less juniors.
for voice acting the tech is already there that is why they are so up in arms. since replacing voice actors can be done today, is already good enough for people to not be able to tell which is which.
How 'fortunate' of you not to talk about VEO 2...
You literally didn't talk at all about the best video generator VEO 2 very dishonest video
i think you missed the additional tidbit that now ai is starting to inbreed itself to death, with previous ai outputs finding their way into datasets (and then in turn causing trained models to hallucinate)
The entire tech space really did just spend the last few years funnelling billions into shitposting technology.
stoncspace
billions to do nothing but whitewash copyright infringment too
Lmao Epic xd
You sir lack imagination/forecasting the future based on recent developments 😉
If you are unable to see the greater picture for the future, I'm sorry for ya. I'm glad that the ones being a little smarter know what the tech can do (and has been doing in the recent years) and are actually able to benefit from it instead of just undermining or ridiculing it.
@@coondog7934>mindless recreation is LE good instead of solving things like resource scarcity and energy
i am 100% convinced Suchir Balaji was murdered. Corporations have been getting bolder and more blatant in recent years in taking out whistleblowers. we need to stand against it.
Perhaps the next Brian Thompson ought to be one of these guys.
If you are a whistleblower and in the US with US citizenship, make sure you use your 2A rights. It may be all that keeps you alive
@@3rdworldgarage450and to have backups everywhere. If it's not in 3 or more places it doesn't exist.
Time to call Luigi fr
_"We hired private investigator and did second autopsy to throw light on cause of death. Private autopsy doesn’t confirm cause of death stated by police._
_Suchir’s apartment was ransacked , sign of struggle in the bathroom and looks like some one hit him in bathroom based on blood spots."_
Poornima Rao's Tweet from a couple days ago.
i've seen a few AI "artist" complaint about their "artwork" getting stolen
talk about irony
Pretty sure it can't even be "stolen" since AI images aren't protected.
@@IamBrainDead Monkey selfie moment
Damn, it's the whole NFT thing all over again, isn't it? "You can't take screenshot of my Bored Ape!!!".
It's sometimes on a level of "stealing" someone's idea, which can sometimes be bad but for the most part doesn't matter much just like somebody making a battle royale shooter because they are popular. Because that's the only thing someone generating an AI image brings to the table, just some general idea, and if it's too specific then the AI can't even replicate it well, and nothing else in the process is a human creation unless you get into the more complicated AI image generation software which 99.9% of people don't do since at that point if you don't have any artistic skills you'll make the results worse. All "creations" are on a spectrum of how much human work went into creating it and for most AI images less than 1% of what matters was human-made in the entire pipeline
They're not artists, they're wannabe businessmen. So eager to circle their life around prompts and buttons. Not a creative thought in their head.
A 'recent' issue that I encounter often is when looking for any references to draw/model/... many of the results are AI generated, but the quality is high enough that I can't tell from looking at the thumbnail, so the workflow is significantly slower than before. AI is not only NOT making being creative easier, it is actively making it harder.
For you mabye, just don't relate that to others., thanks
something i like to do is put "before:2021" in my image searches! kinda sucks we have to do that, but it does help filter out the ai gunk
@@coondog7934 if you are talking about ai making being creative easier, that is fair. Its my experience (I always made all assets myself, so the advantage from ai is mostly getting non-vital code, not exactly a creative thing)
(The other thing you could be refering to)
Going through a feed and trying to find real references is slower no matter how good your perception, just like with ads in a feed. It takes up space and time
At least in Google you can use command “before:YYYY-MM-DD” to search pictures before specific and not specific date
You should type it in the end
For ai it’s “before:2019”
@@coondog7934 god forbid someone share an opinion
As an artist, back when I saw the audio models and the early image models, I accepted the idea that they would end up being very useful for my artistic input, that they would be like "my personal movie studio". Five years later, I have to write "before:2022" everytime I search for references with Google, and the closest thing to "my personal movie studio" is a 2004 videogame.
"2004 videogame". man does SFM/Gmod just refuse to die.
@@ali32bit42 Funny enough I do it with GTA San Andreas (it isn't as advanced as SFM, but it's fun to play with the limited engine). Though a very lazily done machinima does a far better job than AI, not gonna lie.
thanks for the search tip
Use "-ai" when googling, It removes ai results.
@@ibisskb honest to gog, mike and melissa is better than anything ai has ever made
If it at least proves anything, is that we are dogshit at predicting the future
This. Even experts fail A LOT of the time. Most of what we are able to do is several speculations pointing out to too many different directions and then emotionally and mentally prepare for the worst "just in case".
@mekingtiger9095 the more you progress in life, the more you discover that most experts are just experts at seeming like an expert
The AI can't do that either. I send a buncha info to openai, and it tell it "use this to predict the future outcome". It never has the answer.
We want the original design back!! No AI-generated arts, today!!
@@Tall_Order it probably could, it's just programmed to not give you the answer
I forget the exact quote (and whoever originally posted it, sorry) but one of my favourite responses to this whole AI "art" trend was "Instead of doing the tedious tasks for creative people it's doing the creative tasks for tedious people."
When the mining jobs were destroyed in America, I remember when they were all told to learn to code.
People accepted things like that as the natural progression of our society.
Did they think it would stop for them? It didn't stop for us, no matter how hard we fought it.
Instead of helping the artists make more money, it helps non-artists _spend less_ on artists.
This is very obviously the goal, and no one is shy about this.
And while they are not in the wrong, artists can be really entitled bitches about this...
I am a programmer. And while right now, these tools are helping me, I am not an idiot. I also see the writing on the wall. Juniors are hardly getting hired anymore, because of (speculation about) AI.
Right now, we're totally fine. But eventually we won't. Just like artists won't, except a bit sooner.
What is happening to me now is what has been going on with excellent tools like Photoshop, Blender, Unreal Engine etc. for decades now.
Tools that are getting better and better, which let increasingly inexperienced people create "good enough" art.
It should have been obvious for a while now that these tools will reach a point where they will do it by themselves.
Anyone that was caught off guard by this simply wasn't paying attention.
@@hundvd_7 Except that ai can create code too. And the code it creates barely ever executes, because of course it can’t.
If speculation about ai is what is what’s killing every industry, then you know what will kill ai? The fact that the end of the day, these companies all want to sell product. And software cannot buy product. It sure as hell can make “product” but all it does is flood the market.
Bitcoin and NFTs died because they were incomprehensible to most, gen ai is comprehensible, but unexciting. They are trying to sell a niche product and acting like it isn’t niche.
Wow youtube really deleted my entire first comment that would've made this one make sense
@@Mrnotpib Bitcoin price is almost the highest it has ever been, lmao.
One annoying thing about AI, is almost all the AI services charge money for tokens, that are quickly wasted by bad generations that totally ignore several key parts of your prompt. No matter how many brackets you wrap around them. You quickly start to realize that AI is not so intelligent. No chance of a token refund once it's been made...
it's just not a good business model
training and maitaining the models is expensive
and the venture capital is running out faster than it's poring in
@@matheussanthiago9685 the maintaining especially, the resources required to cool the machines that do the computing is astounding.
Adobe took a slightly different approach, it charges for tokens but only after you’ve iterated and approved the end result. If you don’t like it you don’t pay for it. (Adobe Stock’s GenAI feature)
Just set up your own AI locally. No tokens needed, no Internet needed, no mo-ney (except maybe electricity) is needed.
@@coondog7934 Correct. Also electricity is cheap. People have this idea that computers somehow are inefficient. I have heard some absurd takes that AI used the same amount of electricity as the airline industry or something and that is simply not true. They were citing the estimated electricity usage of all datacenters worldwide. So they basically were taking the figure of all the computing power of the internet, and private computer networks combined.
No matter how advanced AI art becomes, it won't stop me from drawing by hand. I've been drawing for a long time.
Real
If anything, it just makes me want to learn how to draw.
and considering the most I can make atm is really bad 2d battle-ships... yeah...
How much do you charge to draw feet?
It will never have context, personal experience, or perspective. It can only regurgitate images. It has no meaning. It can’t create real art, only images. Artists are informed by culture and create culture. Ai is nothing but soulless images. It absolutely disgusts me how every company is forcing it on every damn thing. It makes me even more motivated to become a better person and do better in life in every way.
Same... and one day that will be valued, people will want a work they can hold, that truly is one of a kind (unless a print), something they can see the texture of the implements.
Even if the battle is won against AI the war is never ending. Even if corporations reel back due to backlash, they might try again in a more subtle way until they get what they want without anyone realizing what we lost.
What they want is industrialization of graphic design and other non-art things. Why would I pay someone 30 an hour to design icons for my website buttons when an AI could do it for 5 a month to do it in 10 minutes?
And they will accomplish this. The Fiverr artists who’ve basically been overcharging for simple graphics will be proletarianized like everyone else.
Absolutely. I've seen a video somewhere, can't remember exactly but it made me realise how truly AWFUL the people in power are in their neverending seeking of money. It was about robots in movies. Have you noticed that sympathetic robots, ai and such have been pretty much always symbolic or an allegory to racism and similar? And the bad tended to be a warning. Now I can't remember a recent movie about evil robots, the most recent being the latest terminator (witch mind you, I also don't think is realistic, but I am talking about the perspective of the people making the movies), but there seems to be an uptake of good robots, without the racism allegories and such, in the ways that they are so much less specific and truthful, I guess. It makes sense, when you remember that Hollywood is trying to use this technology.
Think Disney, since they own most of everything anyway, first they put a fake dead man in a not to prominent movie (Rogue One), and we laughed because it was ugly, but like the actor is dead and tHerE WaS nO OtheR wAY...we move on. Same franchise, more prominent series getting traction on the platform they were pushing for. We take it, the actor was still there, and it's better, tHey HaD tO MakE HiM yOuNgER SomEHow...we move on. Small steps they make. Scans of their actors they take, fooling us and them they won't use those without permission...without a contract...without pay...say if there is no contractual obligation and we have the assets, who's to stop us from not paying those actors millions "just" for their faces...ah, the public... So what are the evil conglomerates to do then to make it ok. Media is powerful propaganda and they know it. Kids jump in wells for teletubies, but how subtle can they be for the older people, who don't have much of an opinion on the matter, especially the young generations who don't know the grungy old sci fi and cyberpunk movies, but know a world of greater and greater acceptance.
It is awful, and not all creators are in on this mad scheme, but they will support more creators that they can bend to their needs. It makes me sad, not for the fearmongering of a dark future, but for movies like wall-e, that could tell a story of how humans connect and survive through robot characters, while also having one of the best villains for that exact purpose. Unyielding and unmoving in it's code, not wanting anyone to love and feel and LIVE, knowing nothing else, but following a line. Because it will have to be marinated in the state of this weird war we find ourselves in. Robots were a fascinating storytelling tool, that now has to carry so many more tones and baggage because of all this.
@@gmodrules123456789
Why would I pay even a penny to a robot whose only works have no originality and are simply stolen?
Why would I pay 30 to an artist? Of course, it’s because they’re human, and their works have originality, even if they - in some way - are inspired by other works. I don’t care how bad it is, I only care if the artist is human.
People like you need to stop glorifying artificial -intelligence- stupidity because in the end, it’ll all fall. Art has existed for thousands of years, and your simple underperformance in them makes you seek that sweet sweet little machine to write in a prompt and expect something good.
Commission an ACTUAL artist for that 30 you said, and then give the prompt you gave them to a machine, and compare. Which one is better? Of course it would be the artist because there is genuine emotion put in it. The “art” the machine pumps out has no emotion nor originality, and you only glorify it because you are too snobby and lethargic to even check an actual artist’s works, and commission them.
“It’s too expensive” so is the AI?
Like the Xbox having always online DRM in 2013
Now look at us
@@zenniththefolf4888 UA-cam deleted my comment..
in retrospect, maybe we should have just learned from mario maker... "just because you give people the tools, doesnt give them any good ideas"
Huh?
@@mekingtiger9095 People are really bad at making mario levels
Strong Bad wants to have a word
@@Ballslover18 Ah.
One billion hammer bros at the entrance of the level @@Ballslover18
The additional eyes on the back of the head of the cat unlocked a rare emotion in me that's akin to the Uncanny Valley. I'm pretty sure this is what it feels like to see something truly alien and frightening.
I can't tell you how many times I've seen AI-generated video spontaneously descend into a surrealistic nightmarescape. Its almost like there are malefic, otherworldly entities that are trying to come into our reality, and the prompts are akin to Goetic sigils...
I don't like body horror but I like to use AI to make pictures for fun so I can only make it short burst. So many weird realistic body horror images.
If you look up gymnast or break dancer videos for Open AI and body physics will look really weird somewhat funny also really weird.
I don’t know why, but seeing those eyes in the back of the kittens head made me want to throw up.
because maybe, it is another form of understanding that we don't understand "not everything that sounds like nonsense, has to be a crazyness"
shit reminded me of a spider it was creepy af
I miss back in 2022
Where ai was just a silly little thing that made stupid looking funny images
Then a year later, it was everywhere
And I wish it would just go away, it’s ruining and overtaking everything I enjoy
Me too, man. I remember that I quite enjoyed using websites Craiyon and that website that made AI generated pokémon who's name escapes me right now to make some funny, silly creations. It was fun for what I thought it was at the time.
Safe to say my opinion on such websites have... _deeply soured,_ let's say, since then.
I know better now. I wasn't making anything.
Never thought I'd ever hear someone say they missed 2022. 😂
I miss 'sopping wet beast filling the living room with puddles'
The number of scammers and axis of evil psyops is just stupid. The slop is just too much.
Yes; music, art, TV, film, buying something important, passion projects etc....
Why tf is no one in the comments mentioning how a whistleblower in Open AI got assassinated?
There are some. Just few of them.
Fr
they also got assassinated
Breaking News, Internet user known as Ultimate_Hater75 found dead with 12 bullets in his head, police rule suicide.
RIP Suchir Balaji, the OpenAI Whistleblower
We come at the era of whistleblowers but FTC and SEC had to resort to mooch off CIA's drug money to fund witnesses protection while the tax money goes to Bitcoin, Microsoft, Lockheed Martin, Boeing, and Elon Musk.
It's hilarious how easily people have learned to spot AI. My parents are in their 80s and are computer illiterate, but even they will point at some image on a website and say 'Yuck, AI image'. 1 year ago nobody had this skill.
As for your final conclusions, it's waaaaay too early to say where this will go, it's only been like 18 months. I don't believe the industry in anywhere even close to AGI, I think that's way more complex than they realise. But LLM generative AI will get a lot better, the real problem is that it uses so much resources they have to try and make it profitable right now. But it's currently worthless in all but a few niche applications.
Even for stock footage it's a distraction. But in ten years or more... that's when we will declare this dead or alive. Right now is just waaay too early.
I think people will always be able to tell ai generated content, but it's sad to think that in 100 years, people will think it's normal.
These kinds of technologies are always an arms race.
Cinematic special effects that amazed audiences in the 80s look comedic today. So artists kept pushing the envelope until a new generation could suspend their disbelief.
I suspect the same thing will be true AI. People will look back on these early efforts as cartoonishly bad, but the tools available in their time will seem like magic to us.
Agreed about it being too early. The people writing it off as dead this early seem a little desperate.
agi is a ridiculous goal, and any people that think we're close seriously need to go touch grass, observe the ants on it, and think about how even tiny creatures are so complex
Survivorship Bias.
agi this year. buckle up.
We need to do a better job at helping whistle blowers
For real, companies are borderline openly murdering whistleblowers now. It seems like its all whistleblowers now turn up dead. Corps are hiding it at all, and it's a depressingly common knowledge fact that, yeah, they were probably assassinated
Us? No, that's the government's job because they made it harder in the first place. Espionage Act
And letting the companies do whatever they want surely doesn't help that either
They need to be guaranteed physical as well as legal safety by regulators
Seriously dude, this is probably why many people won't speak out about stuff they have inside info on. The risk of committing suicide with a bullet in the back of your head is just too high. Shits crazy, and it happens way too often to not realize there's an issue.
@@coygusregulators have motive to let it happen
18:31 I've seen people TRY to actually use AI to generate pixel art, but the fact that the fundamental structure is flawed, just makes it so they can NEVER even get CLOSE to legitimate pixel art.
Now I'm really worried because every time Tyler predicts something the complete opposite happens
Dear Tyler,
please predict that I won't find 100$ tomorrow.
new xbox incoming eh? microsoft will get rid of windows and replace it with xbox
@@JimmyCarter1976kinda hard to do that when you’re dead
I think something ai slop has made me realize about real Art is that it is a form of communication. Every piece of art says something about its creator. Every painting, movie, show, book, comic, photo, piece of music, cave painting, napkin doodle, etc. even if its unintentional theres always something that is being said.
We as a species have been creating art for as long as we have been talking to each other, I think that are abilities to communicate and create are the core part that separates us from other animals. Ai has none of that, it has nothing to say, it is fundamentally anti human.
AI art is the same. What it generates isn't a reflection on the AI, but on the prompter. AI isn't just there producing artworks. Someone had to tell it what to generate.
@@DeadFishFactory The issue is that depending on the prompt or the direct interference the user had on the output, the influence he had beyond just a basic conept is so minimal that "the message" might aswell be attributed to the AI itself since it takes far too many creative liberties separately from the user.
@@DeadFishFactory it's not, though. Two artists can have the same idea but wildly different ways to express it. AI "artists" will only be able to make the same generic slop
@@mekingtiger9095 If your prompt is generic, the AI will make it generic as well. The same generic prompt can produce wildly different results because it's starting from different seeds. There's no message and it's basically just statistical noise.
@@LineOfThy It's all in the prompt. Two artists with the same idea will likely describe it differently to the AI, producing different artworks. Obviously, if both are using the same LoRAs or even sharing prompts, they can arrive to similar AI outputs, unlike their artistic abilities which can vary wildly.
Two people with the same idea will produce two different literary works as well. They won't exactly describe it the same way in a poem or writing, and that difference is all that is needed in the prompting for the AI to produce different works on the same idea.
I think I'm slightly less concerned about AI art than I am about AI writing. It's already becoming incredibly difficult to find information on niche topics without stumbling across generated articles with quite frequent factual errors, especially in the tech sector (which feels pretty ironic), as well as the capacity for AI to streamline scamming operations through TTS, which is getting increasingly convincing. All in all, it's just a ludicrous amount of resources dumped into something that i feel has yet to demonstrate a positive reason for it's existence and insane amount of investment - a while ago all this tech was conceptualized as a way to save us time and money, to enable us to enjoy doing more of the human stuff, and it hasn't really accomplished any of that yet - ESPECIALLY not for anyone of lower or middle class.
I think the only genuinely harmful thing AI has done is pollute search engines. My parents and even my somewhat tech savvy friends frequently quote Gemini AI from google because they don't realize it's an AI result.
Literally a week ago a family member who is in his early 30s thought Kathy Bates was dead because he googled it and the first result was "Kathy Bates died in 2022" by Gemini.
7:12 ...huh. I mean, it's obvious that AI does scrape copyrighted art, but wow, that's pretty brazen. Not to mention that tech companies are only going to gain more power over the course of a few years. Yeah this is not good.
I knew about his death. But I didn't know he was *this* close to officially testifying. Makes the timing all that much more cursed.
Nobody cares about copyrighted data when it is used for image-recognition rather than image-generation.
@@Mecharnie_Dobbs that's because image generation is people making stuff that they then claim is their own (it is not)
Human artists also scrape copyrighted art. It's called "studying."
@@LineOfThy Why is it not theirs?
If I carefully study the works of the art nouveau movement to the point where I can paint new ones, are my paintings my own?
If they are, then why is it different if a computer does it?
It is incredible that OpenAI took the life of a former employee for suing the company and not only did no one find out, but everyone went unpunished.
No evidence, no crime, no sentence. In dubio pro reo.
Thank you for spreading misinformation
@@abdimalikelmi729Corporations aren't your friend
Private investigation also pointed out that the circumstances of his death don't seem to indicate the whistlebower did it to himself... Hmm...
@@mekingtiger9095 That's a lie.
The AI bubble burst can't come fast enough
Maybe the downfall of AI was really just the downfall of NFTs 2.0 all along.
Just far more agonisingly drawn out with all the cryptobros who jumped ship from NFTs to it still being neck deep in the denial phase.
The bigger they are, the harder they'll fall.
everyday I pray to any god out there for this day
It'll burst like 3D animation did in the 80s. Never to be seen again!
IF a burst comes. I don't see that yet since the tech still is very useful in a lot of fields.
@@coondog7934 I COULD see major attempts at AI falling apart but that's a little ways off. Like, imagine if OpenAI had to shutter.. fat chance, I know, but if anything close to that happened, AI interest would probably die down significantly
I had CGWallpapers account and was astounded at the amount of tiny details artists put in their creations like tiny bits of screws and bolts, texts etc. and when I think about the amount of time and dedication they've put through, I become kind of sad that AI has devaluated their work for normal people who don't know the details and differences and perfection of these artists who have spent years to master 3D softwares and say: I can do it too now. the truth is, no, you can't. AI is not perfect like human art and would never be. I'm still always looking for human art and real photographs and never care about AI art seriously
So, you don't care about the actual art, just how much effort it takes?
@aolson1111 thats how most anti-AI screeds read to me. Also the interlude with an artist in CS Lewis' The Great Divorce where he flips out at the revelation that in Heaven, being a great artist doesn't get you attention or accolades for yourself because the the actual purpose of subcreation is glorifying the Ultimate Creator and never yourself.
AI art is trash, but it just unmasks the trash that most contemporary art has become.
@aolson1111 there's something special about knowing some every little part of what you are seeing was simmering a person thought about and intentionally did. Like when you play a video game and go somewhere you think literally nobody would go and the developers hide something funny there... it just wouldn't feel the same if that was just created by a computer and not a human interaction like "oh that's so cool that they thought of that"
@@aolson1111he’s clearly referring to the soul of good human art, that every detail you see both has a reason to exit in the world depicted and looks good. Neither ai can ever truly achieve
@@aolson1111 I said I actually do care about human art, not AI creation
As someone that works in 3D modeling/animation and even went to SIGGRAPH last year to look at the developments, even there I think AI is being overhyped. They had lots of fancy stage demos that made it look really easy and useful for environment building. But when I went to the floor show stuff and tried messing with it myself, the results were far less impressive.
The output models are a topographical mess, similar to a photogrammetry model, which is basically what they are when you get down to it. They're generating 2D representations of the object then creating assumed depth from those images and making a model from that. The result is a mesh that looks kinda alright from just the outside but it's polygons are attrocious and the UV/texture is a wonky mess that makes no sense. And even if you make something more complex (i.e. a monkey on a bicycle) it's gonna come out as one hole piece just in that pose making it's actual use without major overhauling super limited. Whether that overhauling is actually less work than just doing it right from the start is a case-by-case issue, but personally I'd rather just make it myself and build it with the intended use case in mind.
Also, the thing has some major limits in generation that I don't see them solving anytime soon. Biggest one being transparency. Tried to make "A brain in a glass jar" multiple times and it had no way to figure it out. Usually put the brain on or sticking out of the jar cause it can't make something transparent with another object inside it since it's building things out as a single blob mesh.
Dude, you are criticising tech that has basically been invented just a few years ago. Just give it a few more years time and you will not use anything else by then.
BTW regarding the transparency thing: Basically 99 percent of all models are trained with .jpg images and those can by design not hold transparency information. So no wonder you won't get transparency right, you need to use a model that was trained with .png images instead. It is not the techs fault if the users don't know how to use it.
@coondog7934 Like how in a few years flying cars would be the norm? Or self driving cars? Or Crypto Currency? NFT's? Tech bros have a horrible track record with this
Actually no it wasn’t “just invented a few years ago”. That is cope. LLMs have been in development for decades. They just weren’t referred to as “AI” until a few years ago.
@@Hourai With "invented" I mean brought to market. You know what I mean, you are not naive.
@@coondog7934 Transparency in 3D means by actual transparent material generation where the 3D object inside another 3D object can be seen thru.
the 3D Model itself must have a transparent glass or similar material on it where the transparency is showing another 3d model inside.
Imagine you tell AI to generate a 3D model from a 2D picture of a Brain floating in a Glass Jar, it doesn't matter the image format is png or jpg, its just an artwork.
What the AI will do is make the 3D mesh of the Jar but once it will try to make the mesh of the brain inside the 3D jar, it will fail cause its just making it all from a single mesh.
And AI doesn't even understand how Transparency even works in 3D (it works like it does in real life to an extent)
Ai might be a generational nothing burger. All the hype, all the corporations pushing it in my face, all the fears and it all lead to no significant change in my life or how I engage in art. I still watch shows and videos made by humans, I still play video games made by humans, I ignore anything that is made by ai and I talk with ai chat bots just for momentary fun and not as a true companion. It’s all just a nothing burger to me. Ai has never and will never replace my appreciation for real art made by real people to me.
Edit: the nvidia stuff is pretty cool though
Same. AI is just best for the likes of shitposting and that's about it whether it's having some art generator make something born from a night terror or listening to someone's video of a cartoon character saying bad words that wouldn't be seen on a kids show.
@@Chinothebad AI is also good for learning programming, especially the massive undocumented clusterfuck that is the Blender Python API.
Fr dude. My back catalogue of games, books and movies extend far back before AI was a thing. Im literally on a binge of a bunch of PS1 games rn. I'll be dead and gone before I even get around to the AI slop games that may or may not permeate the industry in a few years
The only thing I use AI for is to act as google before it was ruined by sponsored results
@@redblue5140 Not sure that's a good idea, AI is notorious for making shit up because it's largely just guessing what the most likely next word or two is and doesn't actually understand what it's saying/doing bcus it's... y'know, just a program. Or what do you mean by 'act as google'? Sorry if I'm misinterpreting.
You nailed it. Generally speaking, AI generated works are created by finding other works related to the prompt and averaging them together. That's why the output always looks generic. You can only generate original-looking output by coercing the AI to average together works that have never been averaged before. But even then, the lack of control makes it difficult to generate anything really useful. And the more control you give the user, the more work they have to put in, the less they save from using AI.
That is the thing though. for people who use the tools to make things practical which require higher levels of control, it doesn't necessarily mean that they are looking to be faster than say a skilled digital illustrator would be to create the same thing. They are attempting to use wholly different skills to achieve the same result.
I really hate when they try to say this garbage works just like a person, it provably doesn't.
@@DarthT15then how do you find Inspiration? because i think artists arent so different in creating something unique, they look at things and then create something with that inspiration but can that even be a one time thing? or are we just copying one another without realizing
@@Crit-Multiplier Ask any author where they get their inspiration from and you'll get answers like "I was driving in a misty night not able to see a car length ahead." Ideas are cheap, they can come from anything. No idea is so insanely clever that it's a final product in and of itself. Execution is what actually matters. It's the hundreds of hours that go into writing a novel. The decade of practice that happened before working hours on a painting. AI cannot copy that. It can only ever copy an end product.
AI cannot copy the muscle memory a painter has crafted that make their strokes unique. It cannot copy all the mistakes and failures that went into crafting a recipe for a cookbook. It can just guess at output that resembles the final product of all that work.
So when you claim that AI is working like humans, no it isn't nor can it ever.
Again this is NOT how AI art works, can you people stop making up your own explanations for how AI works. The lack of control is correct though.
I watched a church full of boomers get emotional over an absolutely God awful AI generated Christmas short film on Christmas Eve. Uncanny movements, inconsistent style, six fingers, missing feet and legs, etc and the boomers were lapping it up. The boomer will keep this shit afloat.
Conservatives tend to have this weird attraction to AI content for some bizarre reason.
@@mekingtiger9095because capitalism hates art
@@mekingtiger9095 welp, another thing added to my list of reasons why im NOT conservative
@@mekingtiger9095 Generalization, old people just cant tell what art is AI created. This is such a weird statement.
The correlation between lead in the brain and ai enjoyment is 100%
I remember once someone said something along the lines of “ai art allows art to be more accessible to people” and thinking, no. art shouldn’t be accessible to you if your not willing to put in the time or at least the effort to make something.
That's exactly the same bootstrap mentality that conservative boomers have about everything else, though. "Don't work, don't eat" kinda thing. People have the human right to be creative. Expression is a fundamental part of being a human. Whether something is art or not has nothing to do with effort, and effort shouldn't be the metric by which expression is judged.
@@_B_E eh not really, I’m not saying no one should be allowed to make any art unless they put in x ammount of hours and x ammount of effort. I’m not even really upset about using shortcuts. But AI “art” goes beyond just a shortcut it takes out any actual expression or creativity a person might hold. When you’re just going to type a couple words and make a machine spit something out and claim it’s art, instead of trying to make something yourself regardless of one’s talent. That’s when I have a problem.
I'm gonna play Devil's Advocate here, though: Yes, art is and has always been theoretically accessible for everyone willing to grab a pencil and a paper and learn. But you want me to be honest? It depends a lot on *HOW* you train yourself. Basic trial and error is the most inefficient way to learn, taking YEARS to show any progress. Fundamentals aren't often taken seriously or have their importance emphasized enough, and 95% of all artists and guides I've found on the internet are TERRIBLE teachers. I've only started to see an improvement with SOME artists providing better educational material since this whole AI thing blew up and even then it is not widespread. Maybe if that wasn't the case, we wouldn't have so many desperate souls resorting to AI as a crutch for believing they don't have the "jeans" or skmething.
This one's on artists.
"No. Good food shouldn't be accessible to you if your not willing to put in the time or at least the effort to make something."
@@Scrimblytimblo "art shouldn’t be accessible to you if your not willing to put in the time or at least the effort to make something."
this is LITERALLY what you said. You're LITERALLY saying "no one should be allowed to make any art unless they put in x ammount of hours and x ammount of effort.".
You're basically just acting like you didn't say the comment i replied to and then moving the goalposts lmao
I want AI to do all the boring stuff in my life so I can focus on the art.
Fr like do my chores and whatever dont do the fun stuff for me???
those in power do not want you doing that, as it requires a level of critical thinking they do not want you to have
@theX24968Z they want you busy putting in a 50 hour work week just to not be able to afford a home in the future
@@theX24968Z Same reason alcohol is normalized but psychedelics are not
Yes, that’s what it SHOULD be for!!
Every possible way to use AI "art" (it is exactly what art isnt) to do something
1: make a shitpost (stupid thing)
2: pretend to be an artist (bad thing to do)
3: make footage of someone commiting a crime (bad thing to do)
4: create ads without paying an artist (horrible thing to do)
5: create spam bots (horrible thing to do)
6: make content farms without even making anything (horrible thing to do)
7: something that can help y ou learn (difficult thing to do)
8: cheat tests (thats only hurting yourself)
9: misinformation (horrible thing to do)
AI can be mostly used for doing bad things and few good things, so in other words, AI is just Riskware
You forgot p*rn
I know youre worried about youtube filters but there's a 10th use heheh if you know what im saying
@@Xenomorthian mate the super sentai ain’t that good
@@XenomorthianWhich is also not good.
@@Xenomorthian 10: making pr0n (it's dogshit)
I don't think AI art will ever truly replace the creative process, because the only people who want to replace artists with AI are greedy, lazy techbros who have zero passion for art beyond its use as a commercial product, and thus everything they make is this gaudy, unpolished garbage that only resembles something man-made if you glance at it for less than two seconds while scrolling thru social media, just another piece of candy to mindlessly cram down your cake-hole. They don't care if their product is functional or beautiful or presentable, or even if it physically exists; they only care about making as much money as possible with as little cost to them as possible, to just press a button and have a computer spin straw into gold for them.
it cannot replace the creative process because theres nothing inherently creative about it. its essentially like using google to search for something but it generates what youre looking for instead. imagine how retarded it would be if we called people who used google search artists lmao
I tried to use ai in writing a story( An experiment), i swear, I always ended up having to rewrite everything to make it more interesting
@tyujg7495. I have sort of the same experience, but with Photoshop. I sometimes use the Content-Aware fill tool to remove stuff from pictures, and it usually does a decent job, but I always have to touch it up myself to make it look more presentable.
it's a shame if AI even takes over stock footage though, it was a good avenue for creatives to practice their skills and get paid in the process
12:50 The only thing the AI could turn white for the winter wonderland was the Asian model...
Tech is built by people. Shock and horror, people have biases
Humor aside, look no further than facial recognition. Years ago, the less Caucasian you looked, higher chance to not even be picked up as a person
@@RadikAliceyou're trying too hard.
Yeah, that was the first thing I noticed. It's like the meme of computer upscaling turning pixelated Obama into a white man (and that was 5 years ago...). You hated to see it then; you still hate to see it now.
asians look too similar to whites so AI can't differentiate properly
@@RadikAlice Yeah, well known. When all the learning data you put into AI is pictures of white people, that's what it considers to be standard. It's amazing how 30 years ago, the first book about computers I ever got had "GARBAGE IN, GARBAGE OUT" as its first programming warning on page two, caps and bold and everything. Decades later, we're still shovelling garbage into the system and getting garbage out, especially as AIs reiterate on one another.
It's tech bros not understanding the fundamentals of what makes art ART. AI "art" isn't art; art is inherently human. An AI vomiting out what it thinks the prompts mean will never compare to even amateur artists. Part of the appeal of art, at least to me, is the imperfections in the line art, the colors chosen by the artist, and the appreciation of the time invested to learn a skill. The only people I think who will still use AI art in the future are going to be corpos who just need it "good enough" (looking at you Coca-cola) to post slop so they don't have to pay real artists what they are truly worth.
Edit: Damn the tech bros are malding.
There most definitely is a spectrum when it comes to how much any kind of vision was being expressed when using AI tools. Just like with photography, cameras can be used to create art, but that doesn't mean every selfie and jpg in your phone is. Art part of generative AI comes from the human learning the insides and outs of their specific model so the "AI vomit" better matches it. I would compare generative AI to sculpting in many ways. You start with a very rough, general shape and then you chip away at imperfections. When you have exhausted the capabilities of one model, continue the process with a different AI model specialized in different things.
There are plenty of human decisions being made when using generative tools if the user has the will and the capability for it. Photographers do something very similar too where they take tons of pictures, choose the one they like the best and then further enhance it with photoshop etc. This process is nearly 1:1 what one does with generative AI as well. Especially as photoshop itself has changed a ton from what it was 20 years ago with so many automated features.
We are also approaching very much the point where actually well made AI pieces are indistinguishable from digital art to a layman, as the picture has been gone through with 1000% zoom and every weird pixel touched on with some other tool until the human in the loop liked the outcome. Because of this when AI tools get better it also skews the perceptions regarding what "AI art" is supposed to even look like, as the people who along the years get really good at using AI tools slowly fade away from public consciousness as "AI artists" due to the quality of their outputs, while simultaneously the skill floor is lowered as the programs themselves become less technical and more "plug and play", which leads to people who care even less about self expression using the tools.
There's metric tons of nuance with AI art, and the spectrum of people using these tools is just as wide. At the more higher-end I'd say AI art is no less art than photography is. That too could be argued in similar vein to be "finding a cool thing and pressing a button, then claiming you made something kekw." But when you add all the technical aspects to it and take someone who is genuinely passionate about it, it becomes art. Even if your uncle's blurry picture of his car he put no thought into and posted on facebook might not be.
It is Art, not high art but neither is 99% of "art" out there.
AI can generate something based on algorithms, it doesn't understand what it's doing, let alone feel it.
When humans do art, they have an understanding of it and emotions attached to it - when you're happy you create different art than sad even if it'll be the same "prompt" to draw.
@@MidWitPride if you're gonna go down and tweak every single pixel, you might as well draw it yourself(especially since usually the output is ugly anyways)
@@player_8008 Art is self expression via a medium, and if someone likes to do it through one method and not some other, it doesn't contribute to the question whether something is art or not. Art is not defined by the effort it takes. If someone likes the process of generative AI more than doing more traditional digital art, then that's what their self expression is about. One could have endless "Why do X type of art instead of Y?" comparisons regarding methods.
OpenAI is spending 700$k a day running its GPT LLM. Let that sink in...
How much does Facebook spend each day running their website?
@@o_o825 If we calculate a year, that value would be around 255$ million a year for OpenAI just to keep running its LLM. Meta spends around 50$ million running its servers. And it earned around 131$ billion in ads just in 2023.
OpenAI revenue from its users was around 300$ million in the month of August 2024 (Source New York Times). Which would push it around 2.1$ million a year in average.
If anyone don't trust me, search for the values yourself please. The 700$K/day I got from the channel Cold Fusion
There are LLMs that you can run locally on any home computer with the latest GPU.
@@Mecharnie_Dobbs Yes. And it is not being used to create countless bots on twitter, youtube, etc.
@@Mecharnie_Dobbs keyword latest. some RTX 4090 computer just to generate text. and thousends of watts of power
I’m an illustrator and for the first year I was terrified, angered and flustered by ai art, but then I realized something. Creatives develop tastes. A palette of media that influences their creative output. The reason why the majority of AI art output is so banal is because the people abusing it have not developed any taste and interact with media in a rushed encyclopedic way that is counter to making anything of creative value or merit.
The bottom line is that "AI" (LLMs) *can't think. They don't think.*
An algorithm, however complicated, can't mimic the intricacy, creativity or capriciousness of actual, genuine thought.
It was always the next bubble for tech start-up land to dupe investors after "NFTs" crashed, in 3 years I guarantee you will see the amount of AI crap being pushed greatly diminish as companies realize there wasn't actually that much money to be saved or made. The AI stuff that will continue will suddenly become less prominent and more in the background, dedicated to very *very* specialized and simple tasks, which is what they're actually good for.
So what exactly is stopping a multi-billion parameter neural network from thinking? Do you think the human brain is magic and the process of thinking can't be replicated?
@@alainx277there is no reason to assume the current method we are pursuing is how the brain works. It is entirely possible we have made something that doesn't replicate the brain at all.
@@alainx277it is you who believe in magic. LLMs lack the qualification to be considered life, which is the basis for conscious experience. "Emergence" does not com just from complexity, but from specificity. No one can claim that sentience comes from anything but biology and be scientifically accurate. To think that your computer software is alive is believing in "magic".
@@alainx277 The problem is that neither you nor me know how thought actually happens. The human brain is opaque, in a number of ways. How do you replicate a black box?
Sure, you could make the stupid thing more and more complex, add a billion or a trillion parameters, but that doesn't guarantee you're any closer to replicating thought.
What you and a few thousand other people are proposing is the equivalent of firing an increasing amount of shotguns into a dark room to hit a target that may not even be there.
@@alainx277 neural networks are just matrices we’ve tricked into predicting data using matrix multiplication and dot products
saying it thinks is like saying a calculator can think because it can do tedious math problems or that a camera can draw because it can create a image of whats infront of it
image generation programs cannot problem solve, critically think or be creative let alone to the extent as a human
outside of image generation, programs that somewhat problem solve like atlas’s ai created by boston dynamics, is still very much limited in terms of problem solving and isn’t designed for complex problems
>2022: haha dall-e is so funny it can't draw very well brrrrr
>2023: AI art is going to bring the end of society and take away all of our jobs. Hyperconsumerism is imminent, late capitalism has forced us to discard our most human qualities in order to buy more. We are worth little as people in this system without opportunities. If our ability to create is being taken away, what's next?
>2024: Oh that's, that's another content farm that flooded my feed with AI Shrimp Jesus Slop. It's the fifth bot I've blocked in a day.
I will not be satisfied untill AI can paint my Wahammer minis.
Bet if someone got hold on one of those mechenical arm robot thingys its prolly possible to do this but with craptons of programing and its probbably easier to do by hand
Print and paint them.
I mean, it could. An arm with brush hold and a program written by chat GPT could do it, if badly.
Here Inquisitor, found a heretic using Abominable Intelligence
You are going to bring the Men of Iron down upon us heretic!
the instant i find out something is not made by humans it becomes 100% less impressive, the people around me and those closest to me feel the same. so i don't think it devalues human made art at all.
and honestly that comforts me. we (for the most part) still desire seeing the display of human skill that it takes to make art and i think as long as we keep that opinion we will be ok for a while.
And what if you use AI together with your own skills to create something unique? Would that be 'less impressive' as well? Your opinion seems to be rather one sided. AI is a tool and in the right hands it can do wonders. Just like in the wrong hands it ends with stock footage and sloppy stuff.
@@coondog7934I mean yeah? People still value hand crafted anything over factory made. Both are made by humans, but one has a hell of a lot more effort put into it.
@@coondog7934Yes. It would still be less impressive. Like riding with training wheels.
@@coondog7934Yes. AI requires no skill
Most of the AI generated stuff shown in this video does not resemble art. It's near photorealistic footage.
I like the idea that at the end of everything that a corporation has gone through in order to cut costs by cutting employees, they'll just have the CEO left as the only employee and he is sitting in a board meeting with the shareholders and they say that they need to cut costs further and do a round of layoffs, then they just look at the CEO and go like, "Thanks John for your service to our company, but we will need you to fire yourself"
I praise AI in one thing only... It actually encouraged me to pick up a pencil and draw. I always liked art but thanks to ai artists acting like the most annoying people i would not like to meet. I just picked the pencil to see if its that much of a struggle and noticed that its rather relaxing to draw even if its just random shapes to practice drawing bigger things.
Task failed successfully.
yooo congrats !! drawing is amazing, hope you keep enjoying your artistic journey!!
AI is so useless. All the problems it could ever possibly claim to solve were already solved by things like templates for writing, calendars on your phone, alarms on your phone, online interest groups, a huge amount of tools in art programs, brush packs you can buy for less than a dollar or even get for free-- all those things already existed and have been used and praised by artists the world over. AI programs demands you pay hundreds of dollars for something that does all that significantly worse. Why would I want that? I've been saying this is the same as the NFT craze for years, because it is. We know exactly how this is going to go, we already know the lifecycle. It's new, come corpopainthuffer liked the sound of the pitch, it's uselessly pushed into things it doesn't make sense to push it into, the scammers get it, and death.
Also the internet is full of the most degenerate, hardheaded, incompetent kinds of people humanity has to offer, and you're training something that cannot read or see or comprehend off their contributions. Ask the chatbots they call AI how many Rs are in something like grooseberry. 999 times out of 1000 it'll tell you there's only 2 because people who were having a brain fart have asked about the berry part so many times on the internet. I used to illustrate that with the word strawberry, but then people kept correcting it so now it sees the full sentence as it's own thing and pegs that specific instance of symbols to mean three. If that make sense. Either way, you can't even trust it to help you in the most watered-down, innocent instances. It's not going to help you with your grammar as an example. more people having trouble with grammar are talking about it specifically with specific kinds of questions than people who can offer flawless advise are offering specific kinds of answers with specific ways to be the most correct. So you got more people asking: "would I use a semicolon here?" than there actually is a solid, repeatable answers to that question. Because that answer changes based on the sentence the questioner is actually trying to write, but the question is the same regardless, you are more likely to get bad advise than good. The AI is going to pick an answer that might not apply to you because it has no idea what the rules actually are, it knows what the most common flags are plopped down after a specific series of flags.
And art is the same kind of flag-flag-raised situation. Your generated imagine of Scooby Doo has his schlong out because the most common training response to low lighting and Scooby Doo come from terrible corners and it doesn't get that you wanted a film noir Scooby Doo, you are not interested in dogs lipsticking.
Funny, i have paid exactly 0 dollars and have gotten tremendous value out of AI tools so far.
@@_B_E All you've done is admit that you're so ignorant that the worst, most incorrect answer legitimately sounds like it is feasible to you.
@@YourWaywardDestiny You seem like you've got a lot of personal things to sort through bud. It's clear that you're blaming AI as a scapegoat for whatever personal failings you seem to be avoiding. I suggest you find a healthier avenue for working through all that.
@@_B_E I'm sorry you don't understand anything at all
@@YourWaywardDestiny Prepare to get your brain fried with the capability jumps this year. I read your long winded piece of nothing comment and it's clearly you that has no idea what they're talking about. AI is already much smarter than you. The r thing is a tokenizer issue and not the dumb shit you're talking about. It's also an expert translation system that is flawless at grammar in many languages.
Haven't watched the whole video yet, but your fundamental error in your previous argument was that you were looking at art from perspective of profit, saving time and making money
To be fair, thats kinda what defines the industry for techbros and cryptobros which are the ones pushing the most for AI art
@@mrjaman3752 "Efficiency, efficiency, efficiency, efficiency"
Tech/AI Bros see art as a mere "market product" to be commodified and not as, well, art.
Reminds me of Zoe Bee's video essay of "In Defense of Inefficiency". Really good one. There's more to this world than just minimizing input for the pure sake of it.
Mentioned techbros and cryptobros shart out a borderline useless product/service like juicero. Only then they try to justify their thing by making up a problem that their thing is supposed to solve. None of the ai/crypto/techbros would ever understand that artists enjoy the process of creating art and there is no problem to solve for the artists. Artis wouldn't ever want something that generates "art" for them.
What so many AI users don't get I that Artists have a deep love and appreciation for the artistic process, not just the end result. AI will never replace us.
You are dangerously naive about the world. Profit and efficiency are what drives incentive. Maybe you should watch cartoons for children
have people forgotten that art is a form of expression and not just a mean to a product?
The Sistine Chapel and Mona Lisa were both commissions. Art frequently _is_ a product.
@@joshwenn989 Mona Lisa was made in 1503 that isn't recent to our time at all. Most art I see is projects or personal art.
@@joshwenn989 Mona Lisa is inferior to the most mediocre of AI art, AI can generate art as beautiful and distinct as the entire Sistine chapel in seconds, AI is already where so many of the greatest images ever made are and will soon be if it isn't already the majority of GREAT songs/movies/images.
@@joshwenn989 i’m not saying there isn’t a product by the end of it
i’m contrasting processes of art with mechanised manufacturing
@@Oreoareyumyumand Leonardo Da Vinci’s entire process was unique as an artist.
With human art every brushtroke has meaning, with ai art none of it does, and you can tell.
It will never surmount this.
Does it?
I can think of plenty of art where the artist was just slapping on paint.
There is some human art where fine care was put in down to the last detail but it's in the minority.
lmao that's very bold to assume
Now tell me about the meaning behind every paint splatter in a Jackson Pollock painting or every splash of sick in Millie Brown's vomit paintings.
@@peterhoulihan9766 What he means is the only answer on why an AI does anything is because it's what it was programmed to. There is no thought process, no idea, no perspective that's being communicated. That's a big deal if you want to avoid being generic.
@@joshwenn989 just because we don't know it doesn't mean it's not there. there is a story to every piece of art to what a human artist puts in. perhaps the artist was frustrated, had bad brushes, really wanted some color in a certain place, or just didn't care and wanted to fill the space with something. it's a nice time to look and appreciate art and try to figure out why the human did what they did. part of the art is not just the result of what is creation, but how it was created, and there is potential for a lot of appreciation for the results from unorthodox techniques like vomiting to produce something that is, subjectively, pleasing to one's eye. even if it isn't pleasing, it can be impressive, and if you don't think so, i'd challenge you to do that technique and show us the result if it's so unimpressive.
one person i'd like to point out that produces fairly pleasing art in a fantastic way is tomwellywells (i've primarily found him on tiktok). his creations utilize spray paint and objects that cover the canvas in interesting ways that come out to producing very interesting looking planets, and before seeing him, i had no idea one could make planetscapes like that.
i am sure if someone knew the technical reasons why llms do what they do, and the data that an llm was trained on, perhaps there might be a similar experience with regards to figuring out why something was created in the way it was, or if they happen to be familiar with a lot of artists throughout the world, if they can find where certain artists' techniques broke through.
yet, once i know something is generated by the llm, even if i thought it was created by a human at first, i immediately start to dislike it, even if the result is more pleasing than some humans' results of art, because to me, the story of the process that drove it to where i could possibly see it carries a weight that text-to-image generation or image-to-image generation simply cannot carry. in the same way that watching something live and watching a recording of something that was live carries a different weight, human-created art and llm-generated mimicry has an ineffable difference that is meaningful to me.
with llm generated mimicry, all these meanings from the human-created art it was trained on blend together to the things the trainers of the model selected for among millions or even billions of pieces of art, devaluing the stories behind them while wasting untold amounts of power primarily driven by a motivation of profiting off of devaluation of those stories. you could argue that's the meaning behind it, to waste that power in pursuit of profit by producing a generic art-like result, and at this point it's then known and the point has been made. they should stop before they continue to contribute to further ecological decline where profits will no longer matter, but i know they won't unless it becomes unprofitable until then.
honestly, with the advent of llm-generated mimicry where companies are trying to remove the humanity from an already underappreciated profession, it's given me a deeper appreciation of that humanity behind the art, and that is why i don't call llm-generated mimicry art. art requires some creative spark and stories behind every word, stroke, and splatter, and while i won't say it necessarily requires humanity (some fish, notably the male puffer fish, produce pieces that i envy to create myself; some people believe our world to be a divine creation), i will say it requires a creative spark and story or several stories that i primarily associate with the humanity behind art.
The AI's own success also poisoned it, at least in terms of art.
So much AI art was produced from mid 2022 onward that it's since flooded the internet and an AI scrapping the internet is going to pick up ever more AI art
AI poisoning is pretty much a total myth and irrelevant.
Do you think that AIs models just hoover up the entire internet without a single second of someone judging what gets used?
@@joshwenn989 Yes, because there is actual proof that AI has been cannibalizing itself.
Such an amazing society we live in! People work all week, overtime, do the same boring repetitive tasks while AI, a non living being, is being creative, making art, having fun! Truly, peak of human society!
Capitalism, I guess?
Every time AI "art" defenders come out of the woodwork trying to analogize about how it's similar to the human process, they expose their cluelessness about the technology they glaze daily. Those arguments could be meaningful when talking about a hypothetical artificial general intelligence that has long term memory and an ability for abstract thinking which is necessary for imagination and composition (a lack of this is what you describe at 10:55). I don't agree with John Searle (of Chinese Room Argument fame) that this kind of broad artificial intelligence is impossible, but LLMs are not it nor will they ever lead to it. I think LLMs are a dead end in the progress towards such a technology.
Altman just wrote they know how to build it(AGI) today, believe him or not but OpenAI delivered great systems(except maybe Sora but that's hard to deliver to millions of users at scale so this is just the scaled down turbo version) that are already very useful for many domains. Expect very strong AI systems in Q3 this year and ASI in 2027/28. Things are gonna get wild fast. Remember to never use never in this domain. You will be proven wrong very fast. Send o1 to 2015 and people would say it's AGI for sure. Moving goalposts is a fun game until there are not any left to move.
@@therealOXOC Altman has been saying variations of that since the very start. It's not backed up by AI researchers from third party organizations or academia, so I have no reason to think it's anything but investor bait.
I have no idea why you say I'm moving any goal posts either, as if you knew me. I have been saying what I wrote above for years and I don't see any serious sign that I will be suddenly proven wrong in the near future. Or if you mean that by some mysterious standard o1 would count as AGI a few years ago, that's nonsense. It's not about relative performance, o1 is simply not a general intelligence. Also artificial super-intelligence within 3 years is straight up delusional.
@@therealOXOC Altman specifically said that they now will consider any model that can make 10 billion dollars of profit AGI, regardless of its output/inner working. If anything, _he's_ the one moving goal posts here.
Nice. I was one of those people who brought these types of problems up with you on the original video, so I'm glad to see the general consensus/doomerism on AI's "inevitable" take-over shift over time. The more publically accessible a lot of these tools become, the more duped the public realizes they've been over the past years, it seems--and the more tired of it they are in general.
I really hope even artificially generated stock footage doesn't take off, I've seen some you-tubers use it and it looks horrendous.
As someone who makes androids, AI is vital to my work and acheiving my childhood dream.
As someone who loves art and likes to see work from talented people I admire, I have a burning hatred for it.
May the first victims of the robot apocalypse be those who exploited and abused it
Unfortunately it’ll probably come for the rest of us as first impressions stick.
@@l1ghtd3m0n3well assuming AI can read the whole internet in less than a second by the time we make it sentient, maybe it can tell the difference between exploiters and real human beings
(Maybe finally all of my info probably being shared around the internet by big corpos will finally pay off)
How has AI art prevented even a single talented person from producing artwork?
Every other new artistic tool (photography, screenprinting, CGI etc.) has only made the creative industries stronger and more valuable. I seriously doubt this is going to be different.
@@peterhoulihan9766agreed
I’ll admit, I like robotics and animatronic things. But I’m against AI art (cuz at least there were people involved in making those animatronics!)
I realize this may not be the best argument but I feel double standardish like that sometimes. 😆
Man, you know what I wish for?
If AI means that one person can do the job of 10, it should mean we get ten times as much cool stuff (or get it ten times faster), rather than 1 employed person and 9 unemployed people.
This issue has existed since cotton weaving machines , why pay more people when you can just overwork one because they can technically do it by themselves now
Imagine a world where all the "chores" job is done by AI while you get to be creative
That's kinda how it works. The endless march of industrial progress means that your wage gets you access to so much more cool stuff in 2024 than it did in 1994, let alone 1924.
@@thelight3112 Eh, and yet most people in the United States have to work a lot just to live from paycheck to paycheck. I'm sorry, but when most people's financial lives are this unstable and are so close to being kicked out their homes, suddenly their microwaves don't seem so appealing.
Technology alone doesn't mean much if there isn't *social* progress aswell.
one of the biggest issues with artists I've talked to is people being unable to put what they want to commission into words. entering a text prompt is just the same problem with faster, worse results so most people will probably never get what they want from AI generators. And imo AI generative fill is also useful for quick, small changes with images and there's def more tools that are actually useful. But yeah, it's not impressive to generate 5 seconds inconsistent of video while consuming the power equivalent of a nuclear power plant anymore.
I read this in your voice
The issue with making your society utilitarian is that art has no place in it.
"AI art" is an oxymoron.
Art is an expression of human emotion. A machine can NEVER create that, even if replicated.
nobody gives a fuck
Yawn.
Graphic designer here, the thing you said about AI images in the realm of stock video is spot on. I work with stock image libraries every day and it’s wild how many AI images there are on platforms like Adobe Stock (thank GOD you can opt to exclude them from search results)
As someone's who has been feeling a bit of existential panic about AI in the past couple of weeks, this really helped ease those concerns, at least temporarily. Thank you
So it's not just me who has been going through the same thing within the *same time frame?* Wild.
@@mekingtiger9095 I've been dealing with the same thing for a few years now.
Same here.
Unironically go outside and tackle something outdoors. Go on a hike, build a shed, etc. You cannot change or stop AI, so why bother yourself about it?
imagine getting a panic... from being horribly ableist lmaoooo
20:08 this is exactly it: there's a use case for this type of tech, as a tool for creative people. But the people making the tech - and more importantly _funding_ the tech - are only concerned with cutting out creative people from their operations. They fundamentally don't understand what makes a competent, talented Creative Professional worth keeping on the payroll, and will happily kneecap their entire company just to lay them off.
There’s also the fact that since the whole internet is fully flooded with sewage, there’s nothing left to train the models off of.
The artists have nothing left for AI to steal from.
Nothing left to train on? All the big companies have huge datasets, both public and private, that are not contaminated with AI generated content.
@ I mean if they ever wanted to get new data, the internet is useless for that.
@@mushroomdude123 Yeah that's kinda true
There is a fair probability that artificial intelligence will attain the capacity to utilize its own output for training without any adverse effects to it.
@@MedicinalSquishing lol. lmao even. why do you think that? give me one example where this has come true. And sources please.
14:55 the funny thing about this is that it is only low compute. On high compute it costs thousands of dollars per task, which means that OpenAI spent over a million dollars on the Arc test AND it was tuned for it. Looking at the examples of what o3 failed on the Arc creators blog makes me question the future capabilities of this technology.
that guy got so excited for the chance to bring down open AI, he took a one way trip to heaven to tell his ancestors about it
6:33 no even funnier they were filmed for real then modified by AI to make it look like it was generated by AI 100%
I think the only ones that will use AI art like that are corpos and people who say "well I can do that" to people who have the talent for art
So basically nobodies who deluded themselves in to thinking they're somebodies.
a means to provide the wealthy with talent, while keeping the talented away from wealth
and also beginner artists who are too scared to post their actual art online so they use gen ai to finish their art or to just get attention
@@anomalousanimates yea if this came out when i was a kid i'd definitely be using AI for my art, because i was hella insecure about it.
The purpose of automation isn't to replace workers, but to scare them by telling them it will.
I got to say the Advent of AI generation has made me much more appreciative of amateur artists and so-called Outsider artists who do not have a technical competency because they're the only people who can make something like that
The same way people adore those who drive around in horse carriages instead of cars because they can and know how to although they don't really benefit from it. Thats how it is supposed to be.
@coondog7934 Not even remotely a good analogy
@@zombieoverlord5173 If you disagree you need to provide arguments otherwise you disqualify yourself from any discussion. Thanks.
Until then my example totally holds up.
@@coondog7934 A.I art used to generate images is not making our lives easier and the arguments against it are not because it's new. Art made by A.I will always be bad and it's helping no one.
@@zombieoverlord5173 Au contraire, mon frère. My life (and hundreds of thousands others) already got better with it. It helps me a lot with my work and grants me time to focus on other important things on the project. Quod erad demonstrandum. I told you my example would hold up.
Shame how so many people seem to be catching a deadly case of whistlebloweritis the mortality rate on that seems super grim
It gives people hope, I guess lol. This is the worst AI will ever be, and all of the kinks could be solved as AI gains the ability to improve itself and teach itself with artificially-generated novel data.
@@o_o825Not necessarily. As AI trains itself on AI generated content, it will cannibalize itself. Also, what does your reply have to do with the comment?
Silly human. Synthetic data is a thing and can only get better.
@@o_o825 Problem is that AI generated crap is always a little worse than the original, if AI can generate images that are 90% as good as the training it and them is trained on that it made it is only 33% as good as the original in 10 generations.
Say what you will about AI content only being a third as good…
No one’s talking about how AI emits far less greenhouse gases on generating content than a human does for the same work. ;)
ai art is like a pleasure cube. It can be just as aesthetically pleasing as a human made artwork, but it will never have the same impact. Real art is meaningful for more than just its looks, the fact that someone thought it out and created it is significant. AI art is like if modern art was actually as simple and meaningless as detractors think it is.
The extra eyes on that kitten scared the fucking shit out of me. I hate this damned technology.
Said by a someone who plays Fortnite, in comparison to which even the most brainless AI generated coco melon doodoo videogame would be better. A bunch of children who are responsible for making such a low effort game so popular that it ruined the whole industry flooding it with low effort garbage games and micro transactions.
Ironic.
Glad to see another Hylics fan standing for genuine art… man, Mason’s one hell of a sculptor
@@tlatai a very inspiring person
@Buzz.co2 And BTW, it’s all fine with me that you play Fortnite. That guy says that Fortnite “ruined” the gaming industry and that AI slop would be better… wish granted! AI tools advance to the point where entire video games can be “created” with a prompt, turning the industry into an endless morass of shovelware the likes of which the world has never seen. And I too pray that this never happens. All these GPUs and gigabytes of RAM, enslaved to crap out another fever dream nightmare. I’ve seen computer programs which are less than half the size of this comment, yet have more infinitely more artistic worth than the next Scarlett Johanssen cabin crew shrimp jesus
When you brought up the thing about meta introducing AI accounts to "boost engagement" my brain seized up
Do you know about robot maids?
It may not completely ruin the professional side, but enjoying art as a hobby (as both creator and appreciator) has never been worse. Having art made by real people hidden underneath gallons and gallons of AI slop is discouraging for both sides (especially when the AI slowly gets good enough to make it more difficult to tell the 2 apart). If this problem only gets worse, perhaps future generations will see art in general in a much less appreciative light, who knows
I’m happy art as a concept is safe, but I’m happier to be able to say I told you so
14:29 also important to note that Google bought 6 or 7 small nuclear reactor to power their AI ventures.
WHAIT WHAT?!
Honestly I'll forgive AI for everything if it makes politicans lobby to actually commit to nuclear power as the future power source.
@@alfonshedstrom9859 Sorry but most AI are smart so they would never even think about suggesting that 😉
@coondog7934 A.I isn't smart. It doesn't even think. Finland is doing quite well with their nuclear powered energy grid btw. You truly know nothing
@@coondog7934 dude nuclear is the best we have, and ai ain’t that smart
I mean corporations don't really care about how generic the product of AI is, I realized that I actually feared a world where people don't really care if something is SOULLESS slop
Knowing how tech bros are it kind of is inevitable that they use computers to compensate for their lack of creativity and humanity. Not saying it's a good thing, but "AI" "art" was inevitable the second we discovered how to synthesize data...
Never underestimate humanities reliance of "Make the robots do ALL the work, so WE...do nothing and get rewarded for it"
AI is not gonna replace artists. People who genuinely value quality in art can see how sloppy AI is. Companies who will change to AI to save money weren´t very interested in treating artists fairly to start with (which might not even happen with the new price hikes)
AI will most likely become a tool to speed up projects, while not fully replacing humans out of the equation.
Looking at how generative AI works (training on existing data to try to imitate it,) you will quickly realize that while AI can generate things, it can't create something genuinely new, it can only attempt to emulate us. You still need humans to make things that have never been seen before. Relying on generative AI to create stuff is essentially pausing the progress of innovation. Also, there is the whole problem of weather generative AI is stealing attribution from creators (which I believe it is.) Additionally, it is very hard to believe all the hype about AI, as things are very over hyped, and the progress of our current methods of doing generative AI seams to be slowing down, and the hope of being able to rely on training AI with it's own output has proved to only lead to the AI being worse. The funny thing about this is that generative AI is often trained on just taking a bunch of content found on the internet, and training it on that, but because generative AI has been so hyped up, there is a lot of content made by AI on the internet, which makes it hard to find training data that doesn't contain content generated by AI.
Essentially, AI will be stuck in the past, it is potentially stealing from creators, it isn't as good as people make it out to be, and it's preventing itself from progressing.
I'm sorry, I couldn't finish this video. The AI videos and images legitimately make me nauseous.
9:45, thought about this some time back, origional art is ispired from life itself. Imagine a made up creature from fiction, it's usually a mashup of other cratures/elements that already exist irl.
It's like the whole 'imagine a colour that doesn't exist' thing. Interesting stuff. I think tripping on shrooms might be a way to create something inspired by nothing.
You make some good points, particularly the way you articulate the issues of control and continuity. But I’m consistently baffled how people mistake current challenges for fundamental limitations. Speak to me of fundamental limitations after 50 years of stagnation and I might believe you’re onto something. After 0 years of stagnation, it’s baseless to claim that.
This key issue as to why the AIs can't produce content you are showcasing is that it's trained to be generic in order to appeal to the greatest number of consumers. That is why when an artist starts to input more content into the prompt the level of intent is never fully matched. There is skill in learning models for their nuances and how to talk to them but it is always specific to the model.
I started leveraging different AI image gens to help me speed up my workflow as I found it gave me the ability to spend the same 10 hours working on fine details of my characters rather than spending 2 hours on details and 8 hours on background plates and poses. As a 15+ year digital artist and graphic designer I welcomed it. BUT, compared to the friends I've made who didn't come from my background, I'm excited to see them exploring their creativity where they previous were unskilled to do so but where I may generate 10 renders to find a suitable source to alter from they will generate 100s and still never feel good about it. In painting helps but so much of business and commissions come down to requirements and intent and AI without post-processing will never replace professionals.
However anyone (other than stock photographers because they did get screwed) who "say never will I ever" is manifesting their career failure.
I believe a good movie can be made with an AI generator but to get there the engineers CAN'T use a general-purpose model and first train a model to build the entire world within it. The characters, the art style, the POIs, the details within POIs. It must have spatial and temporal memory. At that point, you can feed the model a script and let the world play out. It would be a huge undertaking and unfortunately if successful mean a larger explosion of sequels as most of the following movie costs were already covered.
The idea of no one playing instruments, writing stories, painting etc. sounds like hell. Art exists for arts sake, we make art for the love of the game. Fuck generative AI so hard
the terrifying truth is that the generations that grow up in a world where AI is normal likely won't be able to tell the difference, or at best will not care, because they're used to it existing.
If it saves them money and makes them more, corporations will do it, regardless of who it screws over
Great! Another AI video to either make me feel better or deepen my existential dread even further!
I love it!
People who say A.I. art is supposed to replace art fundamentally do not understand what art is. Art is a form of human expression, always has been. Art is used to tell a story between humans. A machine is not going to tell a human story to humans. A.I. will try and find an assimilation of that, but it will never understand how to convey that. Art has to be crafted, art has to communicate. Art is made with a deliberate purpose in mind. In the end, art is made to tell a story(be it something personal, or even just an advertisement). A machine can't understand that, so it just makes copies.
Guess what, noone really gives... The end result is what matters and if that looks at least close to human art then thats all we need. It is going to look a lot better in the next few years.
@@coondog7934 All we need? How about want? How about desire? How about actual taste and style? It does matter. We're not all eating vanilla ice cream, my dude. We have discerning tastes. No machine is going to figure out taste or styles or wants, and neither are the techbros programming them.
@@rawjawbone You are absolutely right but still, who cares? Maybe you do, but the masses certainly don't. And noone takes painting away from you, no matter the technology. So you can still follow that passion as a hobby. If you want to look at human made art you will always be able to do so, since there will always be hobby artists left. For the big majority (and eveyrthing monetary related) AI generation will do for the most part (YT Videos, Games, short movies and maybe even entire full length movies and so on in the future).
@@coondog7934great can’t wait to play a game with 0 soul made by a big company because they need to save money so the ceo can buy their 5021th mansion and not pay their workers a cent
You fundamentally do not understand incentives in the world and what normal people think about AI. Industry is driven by profit, not integrity. Your worldview is childish and annoying
It doesn't matter how good AI gets, that will never stop me from writing. I write because I like it. Having a machine generate literature would be pointless, because the whole point is that I enjoy making it myself.
Exactly. No one is forcing anyone else to use AI.
I use it to research for my world building and to evaluate early ideas. Having someone that can criticize my messy ideas without them stealing them or laughing at me is invaluable.
Add to that a free "expert" consultant that can save me hours of scientific research for my SciFi and I'm sold.
Actual writing is on my hands.
If I ever let AI write something it would be the dialog of one unique type of character: an AI
XD
7:34 sounds like open ai definitely stole copyrighted material
Of course they did but they needed to. Without that they could have never developed their AI since it needs a lot of training data (and buying all those data instead they could have never afforded). And if artists willingly put their work out there for free for everyone to see they shouldn't be confused someone using it.
@@coondog7934 if i make a work of art and show it to you, you cant just steal it and call it your own. copyright is a thing that exists and i feel is often overlooked in conversations about AI.
👁️👁️ 🤫
@@coondog7934 they didn't need to do anything
@@coondog7934 There is an obvious difference between displaying your art for the general public to see, and having that work used without your permission for a product.
What irritates me so much about AI above all else, is that it had potential to be an artistic tool. Stable Diffusion Deforum, DiscoDiffusion, VqGAN+CLIP, they have all been forgotten in favor of pseudo-photorealistic generic crap generators and shitpost machines. The videos made by Linkin Park, Architects, Die Antwoord, Periphery, etc. were masterpieces in using AI as an expressive tool. A Doctor Strange-style superhero film with AI-made psychedelic magic FX made with frame-by-frame img2img StableDiffusion or VqGAN+CLIP would have honestly revolutionised many artistic fronts without costing trillions of dollars and without displacing any jobs or generating any controversy.
BUT NO! It had to become the new iteration of the laziness feedback loop of pointless nothing our culture is right now.
The main issue with such tools is patents, if someone invents a cool new way to use A.I. as a tool they will patent it, sit on the patent for 20 years, and only after 20 years we'll see the technology come to fruition. Kind of like how we could've had a 3D printer revolution in the 1990's, but the inventor refused to do anything with his parents and made extremely expensive and inefficient machines.
Technology would unironically leap decades ahead if countries would adopt a "use it or lose it" mentality to patents, unfortunately, most international treaties explicitly forbid this.
Out of all the things we could automate, we tried to automate art. What next? Robots that watch movies or enjoy meals for us so we don't have to? The premise of AI art was nonsense in the first place.
because they cant say the real premise: "we do not want artists to be able to make a living because their skills to critically think are too dangerous in today's world"
As much as I was interested in AI a while back as well, the fact that I don't touch the stuff anymore is a sign that it won't be as helpful as I initially thought.
I mean, I have used it to (partially) create the character in my profile picture. But since what I was looking for was a different shape for my head and stuff, I didn't need to use it any farther than this.
Every other time I've used it, it was to generate stuff for self-interest, or to get an image that I want to see. Then again, I'd rather look for stuff other people made, just because they are actually refined/crude enough to feel something.
that make's me a little bit hopeful for the future of human artists
No human work has ever been original. Imagination is just combining things we already know
As someone looking to get into the game industry as a 3D artist, the future of AI potentially replacing the skill sets is pretty scary. In theory, AI can be a great assistant in cleaning models, or even handling retopology is attractive, but seeing AI just make a model worries me.
The issue with AI generation of 3D models is that it also kind of sucks.
The mesh is all over the place, it is insanely unoptimized due to a stupidly large amount of polygons, etc.
Like it seems useful but if you had to use it in a game you'd have to change so much of it you'd be better off making it from scratch.
And for how long do we generate 3D models? Exactly, at max for a few years now. You do realize that tech like that needs a lot more years to become a lot better before reaching perfection, right?
@coondog7934 Yeah and we were supposed to already have reached perfection with image and video generation, yet despite the advancements every single fucking AI image looks the exact same and has to get a billion corrections done by humans to fix.
There's only so many times techbros can say "DUDE, just give the technology more years bro please you don't get it." before I get fucking tired.
"Why would someone spend tens of thousands of hours learning to do art"
The way this is phrased makes it sound like the only reason to do art is for profit or social media likes
Exactly. AI is not taking art away from you or me. If someone wants to follow that passion, just go for it and stop blaming technology. It still is a really neat and fulfilling hobby.
The unfortunate truth is that most people tend to approach any sort of constructive hobby as a side hustle.
He meant professionally. art is a dream job of mine. I want to do art for a living because i love it and ai is taking away my dream job and pushing me into jobs i'd fucking hate doing.
I think you are right about the control limitations of Generative AI not being technical. The Tech Bro view is that inevitably as the tech improves the control problem will be solved. But this view ignores a basic conceptual flaw in the entire enterprise which is simply this- people using generative AI don't WANT control- they want the AI to do the job for them. That is-after all- the entire point of using AI.
So to aspire to be an 'AI Artist' is to be in self contradiction- no real Artist wants to hand control over their art to someone or something else- they want to do it themsleves- they WANT control. So when someone says they are or wish to be an 'AI Artist' what they mean is that they like the idea of being an 'Artist' but are not really interested in mastering the skills and knowledge required to actually become an artist.
Certainly it's true to say that there are many 'solutions' available to the AI Artist to try to square this circle- we have 'image2image', or 'inpainting' or 'style transfer' ect- a growing array of tools and techniques that attempt to solve the self contradictory problem of how you both give up control to the AI while at the same time impose control on the AI.
But this is all nonsense- if the AI Artist really wants control the solution is simple- purchase a tablet and stylus and learn how to draw. However this is not what they want either- this is TOO MUCH control! The fantasy seems to be that somewhere there is or will be a 'sweet spot'- a goldilocks zone wherein the level of control will be not too little and not to great- that the technology will evolve to the point where 'something magical happens' and this technolgy will somehow do the impossible by granting granular control over a self guided autonomus machine.
It's this logical absurdity that lies at the root of generative AI. We want machines that offer us total control without the need for us to exercise that control. The perfect AI is the one that does exactly what we want without being guided by us to do it- but how then does it know what we want?
Language-it seems- is the answer. People used to beleive that certain combinations of words had the power to change reality- we called them 'magic spells'. Most people no longer beleive in magic- but we now have a new kind of magic spell- it's called 'The Prompt'- and in place of wizards we have 'prompt engineers' whose arcane knowledge of AI allows them to call forth from these machines all manner of wonders on demand- at least that's the story we are being sold.
But there's a problem. If words could really describe Art in a complete and granular way then reading a description of a picture would be the same experience as actually seeing it- but this is not the case. The reason a 'picture is worth a thousand words' is because words cannot really express visual experience. For the same reason you can't drink the word 'water' you can't really 'see' a picture just by reading a written description of it- you have to actually see it.
But if this is true then how can we tell our AI art generator what we want it to make? If words can't really express pictures how can our prompt engineer communicate to the AI the exact picture he wants it to make by using words? He can't.
The 'control problem' of generatve AI is a linguistic illusion rooted in the idea that words really can define reality- and in any case a real solution already exists. There was no shortage of skilled Artists before 'AI Art' came along- nor was there a shortage of musicians, or writers or any other creative professional- people like doing this work and many are very good at doing it. Generative AI is fake solution in search of a real problem to solve- which is why Meta are going to turn facebook into a wasteland of bots talking to bots talking to bots- not because this is a good idea- even for them it's a very bad idea- but because they are desperate to somehow prove that generative AI has some kind of actual value, that it's more than just a vast macguffin being used to extract money from investors to keep the bubble growing.
At the speed it’s advancing just give it a decade and we’ll see ai really do something special.
The only ai art I want to ever see again in my life is a sentient robots landscape portrait