Task failed successfully

If only there was a Public Domain of art to train AI on, if only!

thank fuck for whoever did that

The data can't be traced or leaked. But the conversation log and the list is not on the ai model. Similar to crypto. The gov don't know who and where it is to. But they can tap your phone and review all your coms to figure who and where.

Even if it's true, we will find out that 4chan get this message as a challenge.

Yup. There are 2 possible responses to that: either 1: "yes you can, you can turn it off. If you can't turn off just the stealing part, you turn off the entire thing." or 2: "well, we warned you, to turn it off back when you could." So go ahead and tell us: Are you lying, or is it gross negligence? either way, you're guilty

Top tier analysis, really went down into the weeds for this one. I think the lawyers at these giant companies are prepared for this reductive line of logic

That litterally just what's happening. The way he represents the situation here is completely false and manipulative. They have yet to proof that what ai does is even stealing. The reason they haven't yet is cause it's simply not stealing. It collects data from images they publically posted. That's not stealing. They use the data to make bew images, that's not stealing. In order for it to be stealing they'd have to be selling or use the originals directly. If you want ai to be stealing than all artists would have to start asking permission and paying for any work they have been inspired by. The basic anime style would be copyrighted. You wouldn't able to use dots for shading or certain line techniques without contacting dc or marvel.

@@group555_ The images are published under specific licenses. Those licenses state what the images may be used for. - Any use must come with credit. and - Training an AI is not part of the rights granted by the license. Are common Using those images *IS* stealing. The thing shown here is: People claiming there is no way to determine what is used as training data and therefore the claim material for which it was stealing, is in any way involved. a.k.a "the loophole" Yet there are leaked internal documents that at least that specific AI-provider, did use hand-picked images they knew were in violation of such terms. And have actively tried to obscure this fact.

@user-jn4sw3iw4h the images are still useable under fair use. Especially the way ai goes about it means that if you call that stealing, looking at an image and later on in life making art yourself is stealing. Also that is not the gotcha they are trying to frame it as. Of course they know the artists where they took the analysed images from. What they don't know is where the training data comes from specifically. To put it in very simple turns. You don't know which 2 numbers were used if all you have is the average. They don't know which parts of each image uses what training data and where exactly that data came from and if they wanted to log that, which legally they don't have to, you'd have a list containing 90% of the training images cause all of them have impact. And all of them share where they have impact. It's not a fancy collage. This guy is just very misinformed, knows nothing but pretends he does, or is purposefully misrepresenting the facts.

now define stealing and how it applies here

​​​@@squattingheadsai is incapable of having an imagination, thus when it attempts to replicate how a human makes art it needs a reference database which it can train itself with and this database is full of (illegally used) artwork. EDIT: stealing is taking something that is not your's obviously.

Dont talk to jasse like that

????

Respectfully turn to ash. I have Not and will never associate with those things.

What counts as an AI bro?

​@@RealityRogue if you defend the ai shit then you are.

Isn't there already one being formed?

Of what? People that gave up the privacy part of artistic moral rights by releasing something on a public forum being a pattern sample?

​@@ANDELE3025 1) This is not how copyright law works in the slightest. In American copyright law, if you made something, it's yours (barring being a derivative work that is not dissimilar enough from what it's deriving to hold up in court). Registering a copyright only makes it easier and simpler to win in a court of law. This is especially true for companies, who have many people creating things for them. 2) Does an artist give up their rights by hanging their painting in a gallery? How is the internet different?

@@polter211 Except it does because its part of international copyright regulations which the US, Canada and even a lot of southern American countries signed. And copyright is unrelated to authors moral rights, the first is a commercial right over exact execution of a idea, the second is what allows you to keep a product private and has to be respected as such. By publishing something on a public forum, you waiver that part of the right, which means anyone can observe, reference, take pictures of, etc said art. And yes that is actually part of your moral rights you lose when you let a painting or statue or any other creation be displayed in a public/open gallery as museums and non-private galleries are public forums.

@@polter211In addition to what Polter said. Artists cannot copyright their art style. An art style is too vague to cover by a copyright. There are 7-8 billion people in the world. Chances are more than 1 person will draw similarly to another person. So you can't copyright an art style or civil court would literally break down by the sheer amount of potential lawsuits. The artist(s) are fighting an uphill battle. They'll have to prove that the AI is copying their already made artwork and is re-creating it as a carbon copy, saying it is it's own or if the AI is claiming it's the artist's artwork when it's not. I doubt either of these is the case, having used AI art generator's myself for fun. The AI recreates the ART STYLE, which cannot be owned or be protected under copyright. Chances are... this will be dismissed cause artists are dumbasses. It wouldn't surprise if lawyers have told them "You don't want to do this." and they're just ignoring the legal advice. And likely AI laws will continue to be discussed on how situations like this are to be taken and eventually laws will be put in place that will make these situations a lot clearer in the law. You should never take the first hit when a law doesn't cover something brand new 100% of the time.

Every AI has its own style though, some are better for more realistic creations, some for anime-style, some for fantasy-style etc.etc.

@@Haerunaits not the ai style its a artist's style that the ai is copying the ai works by mimicking the artist

@@joeywalls4724I mean you're right, I just meant that different AI's have different looks to them.

Yeah but AI mimics in the same way the human brain does so​@@joeywalls4724

@@joeywalls4724 if thats the case all art is stealing

Its important to note that its not just MidJourney. Its every single generative-ai model. There are no artists (pictures, writing or actors) who've ever created enough material to fuel these projects. Mostly because giving your work away so that companies can make money off of paying you for less than your worth is the only way these ideas can be actually used..

Nah it's a miracle. Having hard examples of these companies lying through their teeth while steeling from artists is exactly what's needed to have any chance of getting any protection in place for artists. AI art isn't inherently bad but it *cannot* be trained off of art that isn't licensed. Special requests regarding asking something to be done in the style of a particular artist aught to be protected as well. This is at the very least necessary for anything that's used commercially on any level.

@@bwood6337 I very much agree, thanks for the perspective!

​@@bwood6337that's not how copyright works though. The product put out isn’t the same as what was put into it. Transformative is one of the requirements and that's what is being done. Also when it comes to tracing things back that list isn’t what they're talking about. What's being talked about is if you look at an image you can't analyze it and say "this image came from ****" in fact legally speaking as long as it's transformed to that degree as they claim all that list is, is recognition and can't be used against them anymore than tissue paper can

@@tomraineofmagigor3499 I don't think this is a matter of just copyright in the first place. At a basic level these companies are nonconsentually reselling someone else's work on mass. It's doesn't matter if it's transformative. Imagine if the same thing was done to actors in movies/series where one of these large learning models ingested every piece if media from the last three decades? Ovbiously at that scale the end product would appear transformative, but that doesn't make it any more acceptable/ethical.

It doesnt violate it so i dont think you want to go that route

I think you misunderstood, the list is what the Algorithm used, which means it was needed

no i think you misunderstand, that list is a human-made list to track it, you don't feed algorithm that messy data made by humans to understand what the data is. you clean datasets for algorithms, not for humans. That is a list separate to track what data has been fed. and honestly, a list like that should never exist available online for anyone, whoever input and tracked the data messed up big time. But you don't use .txt files with that much junk in them as a dataset, head over to kaggle if you wanna take a look what a dataset looks like@@candlestone5397

It's not a crime. Art styles are not protected by copyright.

@@Anbuscythe But art is, and you need to steal art to teach the AI.

​@@Anbuscytheit's not the art style people are complaining about dumbass. It's the data and literally art stole.

Midjourney is an AI. What do you mean? It uses transformers in a deep learning neural network.

@@BlueSapphyreits a deep learning algorithm. AI is a misnomer because it isn’t actually intelligent. Its just an algorithm

@@sashaaurora8983 Deep Learning is a subset of Machine Learning Algorithms, which fall under the AI umbrella.

Then please explain. Because I can feed ai pictures of the style I want, and then get that style returned my prompts. So, it's pretty similar.

@@Nobody-vr5nlNot at all. It can’t create anything wholly original. All its datasets are based on art it scraped. A human creating it however will always be different. In a sense, Learning Algorithms are much more akin to tracing than just an artist drawing in a specific art style

trying to hide artist tags is purely for PR. Scraping isnt stealing by legal terms. You have to sell the protected works directly for copyright protections to kick in. and the content produced on the other side is unique enough from any copyrighted works as to be their own thing, even if they closely resemble a copyrighted work. Ethics aside, legally they really dont have a whole lot to take responsibility for.

Guess I should be in prison for all those drawings of anime and cool fantasy things I traced in middle school.

Problem is it's not stealing and if it was that cut and dry there would be court ruling over it but no lawyer wants to tackle it. Ai is better than artists already and soon will be even better, a lot easier to work with too, concept artists will whine that you're toxic bc they're taking 2 weeks to draw a single concept but they're never working. My machine can give me 100 concepts in an hour and I can upgrade it.

the original quote, I believe, is malice? But I think more accurate here would be *greed*

@@shilohmagic7173 and flipped, malice then ignorance

@@spelcheak It's a very versatile quote format!

That quote is backwards "Never attribute to malice what can be attributed to incompetence" Or if big words don't work "Cockup before conspiracy"

@@minecrafter8001 oh you're right it is incompetence!

Who's benefiting? Well, of course it's people who don't want to pay artists and the people who set up midjourney.

Lazy people who don't want to learn how to draw but still want money.

There are people who think they can't draw, so instead of working on their skills and practicing, they use AI to communicate their ideas.

Because making things worse for money. No other reason. Some ppl just love that shit.

As Mikedsps said, people who don't want to pay for art and Midjourney. I understand people who don't want to pay $30 for a cute cat wearing a wizard hat. (I might be exaggerating here but still)

didnt know they knew eachother lol.

It steals directly from the artists for profit. No one would complain if artists got paid for ai using it

If you think thats the same thing as Ai art i feel sorry for you

When artists do any of that, they usually at least try to credit their references and inspirations. The issue with AI nowadays is that it usually doesn't do that, while taking art from several artists at a time as a basis.

A human can't create art in the way AI generates images. There's no self-awareness involved in the process. How could it be art if there is no artist? Even a person in a fugue state would have some basic connection with their humanity to be able to influence the process.

@@IronTiger Hence why it is a philosophical discussion, what does self awareness mean? Are very young children self aware? Are animals self aware? Where is the actual line and what sets AI systems as truly different? AI systems are designed with a variable known as "noise" - which can allow deviation and emergence away from a fed dataset. Is this similar to the way a human being might move away from pure mimicry to novelty? We also have plenty of artists, writers, musicians, etc that have written books and talked about how using others' works as a starting point can be beneficial to the creative process. I personally don't believe that whatever we call "creativity" or "innovation" is purely human because I see similar problem solving capabilities in our cousins the Chimps, but also other generalists like bears, crows, octopuses, and more. And a crude way to define life on Earth is a combination of really wet rocks that became self aware, and don't see what that capability cannot eventually be replicated in AI systems

@@Barrel4336 I mean self awareness that one has the ability to choose. Machine learning image generators require human input to initiate, augment, and iterate on processes. A human artist can do these things autonomously. We can choose what we make, how we make it, and when it’s complete, and from beginning to end of the process what we intend for the viewer to perceive. The viewer is also able to evaluate art in a way machines cannot. Every artist has a life outside of creating art which, in some way, influences the creation of their art. Even an elephant with a gift for painting is closer to a human artist in that respect. The only part of the discussion I consider philosophical is whether it’s worthwhile to evaluate machine learning generated images as art. I personally feel it is not. If we develop fully autonomous machines that have “lives” outside of a sole function, I would reconsider that. Show me a self-driving Google Streetview capture car that decides to recolor something it photographed purple instead of orange and can give any explanation as to why (“because I felt like it” would suffice) and I’ll consider that AI art.

@@IronTiger The reason I see no difference is mainly because humans and animals are a form of biological robots with our own simulated intelligence (that is in if itself highly incomplete and can be selective in the way perceives the world as it is processed by our minds) And we can be certain biological systems likely had an "beginning" as well regardless of it being emergent or not and the only thing we can guess was certain was a drive to "survive" or at least successfully replicate. All other systems: social relationships, world views, creativity, art, etc. are secondary emergent properties born out of that core drive. Just like we can't be certain that AI systems may eventually transform into something more autonomous with their own emergent properties (which they technically have as machine learning systems are developing their own encoded languages to better process information independent of human intervention), we can't be certain that we are selves are not a product of a similar evolution trajectory in the past.

​@@IronTigerit's crazy to me how every single argument against AI is always it doesn't have a sole. Which means nothing to the majority of people

That's a crime

​@@themechanic9974 And?

@@genericname2747 end of sentence

@@themechanic9974 ok 👍

You mean the thing that doesn't work?

The AI will be killed by the lawsuit.

​@@GermanKinsmenno technology have been successful killed by law

It's not actually AI, it's machine learning. It's equivalent to calling a car a horse because it travel from point a to be with a load capacity. Ai used to refer to computer intelligence that can react independently without human order/interaction. Not a computer program that can only produce what it's been told to do.

@@RandomPerson-tz7wk interesting argument, I agree with the diferentiation but not the reason or the anedocte. Food for thought What is the product after the machine learning? When the learning is interrupted wouldnt be acceptable to call the model AI by most people? What happens if the AI had been given the exact purpose of only responding to people? Would this make it less AI?

​@@gabrielxavier2676 Product of a machine learning are parameters for the model. If you interput it early, you likely get bad parameters.

It is fairly funny seeing people use the dual arguments of "it's just copy and paste of real art" and "look at how ai art has screwed up hands, so it's a terrible thing that doesn't understand art" at the same time though.

Riot Games is rich and absolutely not normal people, and they're already putting together a corporate lawsuit.

@@GermanKinsmen yeah, they’re NOT normal people and thus have nothing to do with what I said. Is that supposed to be a gatcha?

The problem in the AI creators' eyes is the fact that they would have to pay THOUSANDS of artists for their licenses if they wanted enough training data for a decent AI art program. This would cost WAY more money than they could hope to make by using the AI, when the entire original point of AI art is to save money by not having to pay the artists in the first place. In other words, AI art in it's current form is genuinely impossible to create ethically, hence why every AI programmer and corporation, unsurprisingly. decided to just not bother trying. I doubt that everyone who uses AI art is looking to fully replace artists, but the fact is that for EVERY person and company making the AI, that's exactly what they want to do.

@@jonathanflanagan1504 not quite. The real root of the problem is who do you pay? art is derivative by it's very nature and artists have been stealing styles for as long as art has existed. now who do you pay? z who took from x,y, and a, or y who took from x, q, v, or do you go farther down the line? It's an infinite regress. Now since ai must pay why aren't the artists paying a license fee also for doing the same thing? It's a case of special pleading that doesn't equate to the reality of the situation.

No way it's public and fair use. If I learned from a artists public works then recreated their style there is nothing stolen the machine learns like a real person it is not just a program on a single machine. Why would I pay for an inferior product when it's free.

Yep. It isn't "literally" how the brain works, but it very much is a simile when it comes to the process. But artists kind of act like religious people, grasping at any straw and twisting any argument for their benefit because they feel attacked on the emotional level. No need to make up arguments, the case is pretty clear. We need laws that prevent copyrighted works from being scrubbed and used in AI training phase without permission. Then, you can either give that permission for free or licence it out for an amount of money. The thing is, during training the model "reduces" all that material to a set of numbers that are nothing alike it. You can't look at the numbers themselves and say "Ah, John's art was analyzed". You have to look at results, and that's... Even if it spits out something that looks like your style -- maybe among thousands of other art there were some that had similar style to yours. Are you going to now accuse those other artists for stealing your style? What if their art is way older than yours? Did you then just effectively accuse yourself of stealing the style? What if the AI combined styles similar in different ways and the prompt was such that the combination looks kind of like your style? How do you prove your art was analyzed? Companies would need to be transparent about training, because just analyzing results would be an exercise in frustration.

pretty sure that's just the light cause you can see the bottom of his face get lighter too when he moves his head

That's why the generators aren't taken down. They love calling it stealing but they have no actual legal point. If they keep it up they might even be sued themselves for defamation

i have typed 3 different full length essays to say no to this but let me just say this; AI does not copy and paste, it recognizes patterns, just like us.

So you think the AI on your phone is a living person?​@@minecraftnoob-vu3ye

@@minecraftnoob-vu3ye The AI is perfect at this though, whereas a human is not, and fills in the blanks through other sources. The other sources can be literally ANYTHING and not just other art. Because life is endlessly complex and the human experience is quite literally different for absolutely everyone. In order to have an AI that can truly be original, it has to be trained on more than just art, and then it has to be able to express the non-art data IN art. In other words, the AI cannot produce original work because it cannot experience an original life.

@@jonathanflanagan1504 can you state those “other sources”?

@@minecraftnoob-vu3ye Emotions are a big one. Many artists draw what they feel, not just what they see. My point is that when an artist learns to create art, they're filtering it through who they are as a person, their likes dislikes. Art is a form of expression. That's where real originality comes from, from the artist's perspective formed on a foundation of their life experience. AI is utterly incapable of expressing itself at the moment. It cannot put a spin on a concept. It cannot create base on what it feels because it feels nothing. It cannot draw on experience outside of the existing art that's been carefully and maliciously scraped from the internet to feed its data. It can only do, EXACTLY what it's been told, however complex and misunderstood those instructions can often be.

we spent decades on psychology and neurology to understand how the brain works, and then decades more on computer science to simulate that process, only for laymen and artists to remystify the brain claiming "human brains have special sauce" that computers can never have. Its a purely spiritualistic thought grounded outside of science.

AI can ONLY be trained on art, while a human can draw from every aspect of their entire life. I won't call generative AI images 'real art' until the AI making them also get real lives.

@@jonathanflanagan1504 That's not true at all. ML algorithms can be trained on a variety of stimuli in multiple forms.

If you ask a artist, then it isn't. If you ask any rational person it isn't. If you ask an artist thats upset about AI it is... The problem is artists are not software engineers. They do not understand the fuck they are talking about and getting emotionally upset about this and using the wrong arguments. There is an extremely solid legal case to argue that AI training data should only come from artists that approve their work for commercial use and are paid for it as is in line with copyright. But the argument artists keep clinging to that a neural network and humans learn differently when you're talking about source material based learning practices is just flat out wrong and damages their own actual stance. Just as you can copy someone's style and learn from it in a personal project or even professional ones, in many cases, so as can a neural network. There is no ground to stand on there. It's fundamentally an issue of “AI” is being used as a scape goat to circumvent trademark and copyright, and THAT needs to stop. Scape goating needs to be punished and that is what Midjourney is ACTUALLY doing.

​@JustinPerrySparro but human brains and the current generative art programs do learn different. human brains are subjective, machine learning is objective. Saying that its like the "inspiration" that humans use is wrong because inspiration is subjective. Also from what I've seen artist are mostly angry that these tools are plagiarising art. It is with the intent to profit off of them.

​@JustinPerrySparro ai art isnt art. And artists shouldnt be stolen from just bc you think differently

AI isn't compositing. It's literally using probabilities to deduce what pixel goes where, or how the next pass on focusing a mess of static to make it "more like" what it wants should be generated. Pass by pass, step by step, until it reaches the specified amount of steps and weight of the specified prompt. It's really good at understanding and learning style. You can generate a Pokémon comic in the style of Bruce Timm (Batman the Animated Series) and it can just work. Even though Warner Bros. never animated Pokémon. (Except maybe for a bumper on Kids WB lol). It's not compositing or morphing any art of Pikachu or Batman. It just knows the target image (Pikachu in Batman the Animated Series) will follow some of the rules, design, strokes, etc. as a drawing of Pikachu in general, and some of the colors, angles, style, etc. as drawing a scene in Batman the Animated Series. It isn't using source material of either in particular in the final work. It just used some source images to learn how to make them in reverse at one point. Imagine I have a picture of an apple. That's the endpoint. The computer messes up the image of the apple into static, and records how it got to static from the apple so it can reverse it. Then I ask it to turn new static into an apple. It generates a different apple using the same algorithm it learned from the other apple in reverse. Do you own the new image of the Apple when I feed it new garbage because a computer taught itself how to screw up your image into garbage, and how to reverse garbage back into an apple? No.

Lmao no, tracing is directly copying part of an image. Even just looking at what ai makes it's much more similar to taking inspiration from. As in the very thing every artist does

neat. I wanna hear a biochemist talk about how ai works. got a link?

More accurately, it would be like you spending years studying art, learning by comparing and observing millions of different pictures, and gradually getting better at incorporating those ideas into your own work, then creating a custom piece of art for someone based on a short description of what they want

@@4Ninjastarz4 so the user isn't an artist.

@@scw55 That's correct, the user is more like an artist's client while the ai model is the artist

Labor and creativity aren't necessarily required for art, there isn't any when a person presses a button on a camera but we still call it art. Also something being a collage doesn't make it not art, if an artist made a song out of various samples from other songs do you think that could be art?

@@StFido you grossly reduce art photography. Casual snaps is like a casual doodle. An photographer artist has to consider their equipment. Consider lighting. Composition. Consider communication. All creative or technical decisions. Choices a casual snap won't include. It's the difference between a fast food burger and a restaurant meal. Both feed you. Both give enjoyment in different ways. But it's a disservice to both to call both of them the same. If you made a song from many samples, it can be art. But it's certainly going to be hell to monetise unless you get permissions or licences for each piece. However, this is a human process. A human took samples and wove then together. Making numerous micro creative decisions. By misunderstanding my collage description you also dismiss that as an art form. This practice involves many more micro creative choices. Art is communication and making creative choices, informed by practice and study. The images made by the models weren't ever made by you. Legally, they aren't even your's.

That's not what they did. As a side effect it can be used for that but that's not the main goal. The goal is get it to make general art that can shaped by the user. I can use word to copy a novel and sell it as my own, does that mean word is a scummy technology?

um...buddy...theft is what art is. artists are just as "guilty" of the claims being made against ai. They just don't use special pleading.

Your telling me people need evidence to believe something rather than hearsay?

Bro are you people simple? First off, this guy is wrong. Second, what the AI does isn't theft. It's learning. It's no different from a human looking at your art and being inspired by it or imitating it. Only the AI is much better at it. But third .. this dude really say that "the tech bros" don't know what they are talking about .. it's the "artists that really get the tech". foutta here bro

@@grum384 Having a list of artists isn't evidence. Looking at a image/learning from an image is not theft.

​@@3choblast3r4theyre not wrong. Its not fair use, you can tell by how the courts decided they cant use it for commercial purposes explicitly because it is not fair use.

Because artists probably don't know how it works either.

In what way?

You should probably learn the first thing about this topic before saying shit like this... No, its nothing like learning because it doesn't learn. Every single algorithm used needs to have human-idenitified tags that both selected *for* and *against* proper use of a tag. When you google search "cat" you get more images of vaguely outdoor pictures more than cats that are shaped or coloured like each other. So any recognition system would identify the grass to more apart of a cat than a tail.. Because of this fact, all ai-art simple takes mass amount of tagged images and copies the most common part of all of the tagged traits. it is *impossible* for these programs to create something new. Your brain is tricked by the collage effect, the ai cannot learn, it has to be programmed to dedicate a tag to a trait. The ai cannot seek a pattern, it only sees the common pixels related to a tag. the human brain can view something and interpret something from memory, it is faulty and thats why 100 people asked to draw something can come up with very different ideas.. But if you ask ai to draw the same thing, all those ai would come up with the same thing. Factually not learning. You watch too many movies. Take a computer science course

@@TheKiroshi Well I am a software engineer as a profession with over 20 years of experience, so the computer science course is of limited use, but thanks for the suggestion. As far as tagging goes, you're somewhat confusing the concept of creating human interfaces from the input data. Yes images have to be tagged, but that's purely to give a handle through which it can communicate with other humans. And in reality, once the data is processed, it's not even stored directly as tags, instead there's layers of conversions of "this is what this AI understands the token 'Red' to mean, and here's the art definition of a token that happens to be linked to the concept that we think of as 'Red'. And wow, that's pretty much exactly the way humans learn. When taking art courses we're told "this is what impressionism looks like" and "this is what cubism is" or "this is classical art". All through out life we're shown animals and told "this is a wolf, it looks like a dog, but it's different". Humans learn by associating words with abstract concepts and then using those concepts to produce art, solve problems, etc. And that's the problem, humans learn much the same way as this AI does, unfortunately most artists are too blinded by their own biasis, or misinformed so that they think there's major differences in play.

@@halgari -- All you said was "this is how the human brain works" though interpretation than you just ASSUMED that's how AI learns. which it doesn't and you proved it so above. All these have to be tagged. It's not creating something new, like I said before, If you only show AI the picture of the front of a cat, maybe by coicidence, they all sit on their tails.. that model will *NEVER* make it with a tail. Humans intead can look at all the other animals and related things and assume it has a tail or not. Ai cannot do that. it cannot learn, You're just being tricked by the idea that programming is random. "artists are blinded by their bias!" by proof? Literally any proof? Your *Supposedly* a software engineer (which means literally nothing, Could be a programmer, could be an ideas-loser, could be a corperate suit, Could be youtube UI designer) You *SAY* Ai can learn.. But every single model ever continually needs to be given human input to randomize and trick you with the collage effect. AI has never once, ever in history, done something it's programming has not allowed for.. Humans learn via sensory input and interpretation. AI repeats programming. Ai cannot interpretive anything because it cannot think, it don't reflect on anything. Like your bias is to think techonology is some magic state, And that's making you buy into the lie that these companies are selling you. Artists at worst have a bias that replacing art with forgeries is bad, and that every single time it's used in any setting beyond "haha. funny pic" it's literally only for the benefit of a corperate dollar. All you need to do to convience my otherwise is show me the code. Go ahead, "Software engineer of 20 years".. Go ahead, Show me the code that proves AI can think and interprete without human input. *Spoiler alert, you can't*

Corporate law is not cold and unfeeling. It's biased and weighted towards whoever has the most cash/favours to throw around. There's a difference.

Ai doesnt "incorporate it into something new" it cant think for it self. It doesnt come up with ideas for a peice. It just steals already existing ones. Youre coping hard

@@1RiverWang you are objectively incorrect. It sources ideas from multiple places and incorporates it into a new work. That is a fact. Whether it thinks for itself or not is irrelevant to that statement. I'm not the one coping here.

@CGoody564 yeah Ai steals. Thats what youre describing. Ai cant create art cause it isnt human

The issue is labor. Humans can do labor, programs can not.

​@@1RiverWangthe ai cant think but it can't make its own decisions if this wasn't true a artist could replicate ai style but they can't. The ai can create things never thought of proof it's original.

Every time someone had tried to sue over being scrapped for AI training it's failed so probably not.

Something has to do something wrong or illegal for a class action lawsuit to work. Want to know why there has not been any successful lawsuits? Because the people being accused have done nothing wrong.

It doesn't matter if they "cant tell which artist an image took from" if the names on the list without consent or the work isn't paid for they can be taken to court for it and lose to the owner of the IP.

The A.I. is not an artist nor are they a person. It does not “learn” in the way you and I do. It does not develop talent. It is a tool, and it is being powered at the expense of real people’s livelihoods and hobbies. The difference is fundamental.

none of this disproves anything i said though. the end result is still the same if me or an AI reproduce the results. if i learn as a hobby to draw, or if i learn as a hobby to make AI art, and result is a picture at the end, how is the end result fundamentally different? also we dont care if a tool impacts peoples livelihoods. if we did we wouldnt have cars since people use to have jobs pulling carriages on horses. and people also use to draw art of people, and now we have cameras. this idea that ai is bad because it takes a livelihood has the same sound as people being mad at cameras for taking the job of someone who draws portraits, its an insane take since rational people understand both can exist. cameras and artists both capture trees, and yet there are people who take photos of trees, and theres also people who draw trees. both can exist and to act like one destroys the other is crazy

If you learn to draw from someone,you can 100% replicate how they do their art. Sure it can look similar but it’s literally impossible for a human to make an exact replicant of someone’s art . AI doesn’t learn how to draw,it’s programmed in how certain artists do their drawings and takes bits of that info to create art that could be very or fully close to the original. Plus,artists a lot of the times rely on their art to eat,ai does it because it’s told too?

your joking right? you cant actually think humans cant replicate art? first off photos exist. secondly a copy of the mona lisa sold for 300k in 2021. now im not an art expert but i dont think davinci painted that one, so the only other options are someone created a device to clone things (and in our technological world thats getting closer every day) or someone else painted an exact copy of it. which do you think is more likely and can you explain why you think humans cant replicate other things humans were able to do?

​@@blader36850 You're personifying a machine. It fundementally can't learn, because learning is not a thing these programs were ever capable of doing. The process of learning is an extremely complicated process of neurons connecting through chemical stimuli that only a handful creatures on earth are capable of doing to a complex degree, that even the highest forms of research can still barely and vaguely understand. Ai data librimitations (because that is what they are, a library of data it mixes and matches) are not people and therefore could never learn anything, they accrue saved data. We know this cause we built them to do exactly that. Accruing data on a system that can single out the circles where an artist draws eyes or the color the grass they draw is not learning. And to your point of "What's the difference if the end Result is the same" Well if these ai art generators suddenly all corrupted and couldn't generate these collages anymore, we'd still have art. But if all the artists' art was deleted off the web and the artists stopped making art, ai art generators would cease to function, having nothing to scrape off of to create their immitations.

You're talking like you just have no choice. Apparently you *need* character reference sheets and will harm others to get them.

You can just use art for your d&d games.. if you want "hunter with crossbow" you don't need a program that charges you for actual use, or the re-re-re-re-reattempt to get a half recognizable picture. The issue with AI-Art is that those programs pull entire libraries worth of pictures to copy from and often charge you anyways. These are companies using other people's work for profit. You using it for your "haha lul" roll20 pic is not a problem for anyone.

@@MuchToDoAboutNowt the 'harm' here reminds me of video game execs whinging about piracy. no the broke kids were never going to buy your overpriced game, just like hobbyists were never going to commission art for fun. the harm is mostly to your feelings.

Claiming that it's theirs and not from the original is stealing tho!?

Baby has legal grounds for a lawsuit.

Does that mean I can trace the Mona Lisa and sell it as an original piece? Just a thought

As long as you make a original twist the the thing your referencing(which is not copying) or draw it in your own art style then it’s ok,but if your basically making a carbon copy of the original piece or only making minor changes then it’s considered copying.

But ai art is something different as it’s not an actual person making the decisions and is basically making a collage of drawings from multiple different people then I would consider that stealing or copying.

I get what you’re saying, but in practice that would be difficult to implement. Because you don’t use all the data in your training set, the algorithm randomly drops some of the data to prevent overfitting. So replicating would be incredibly difficult. But the model is run millions upon millions of times and an average result is taken, so replication could be close, but rarely, if ever, the same.

And every artist that doesn’t declare their list of influences can be fined. Any law about this will be used to grind artists into the dirt with the same logic.

this wouldn't actually change much. legally, currently, any image you can right click save as you can do whatever you want with it. if you can access it you can keep it to use for other work. images aren't normally used this way but think instead of a look up table put together for, idk, the price of fish in different places. you can do anything you want with that if you manage to find it so long as you didn't do any breaches or fraud to get it. besides, how do you know if they give you the full list and not an incomplete list? it's not falsifiable.

Even if you used the exact same dataset to train 1000 identical models you would get different results from each one

@@4Ninjastarz4 just not true, you can tell them to always get the near exact same result. There is no need to always have it „random“. Rng does not exist in computer science exactly. If you run an AI with the same input for 1000times it is highly likely you get multiple matches.

If AI replaces artists, the AI will have nothing new to train on.

This isn't about "influenceing" art, it's literally theft, they are taking literally pieces of their artworks and mashing it with a bunch of other pieces from a bunch of other artists' artworks.

@@ghostphantom8453 that's what ever artists I've ever seen does

​@@hotrodhunk7389that is not how human brains and making art work. Why do you think AI art is often uncanny, struggling with things like hands? Because AI doesn't know what a hand *is* -- it just has data labeled "hand" that it pulls from.

@@unnamedenemy9 yeah that reminds me a lot of humans when they're first learning to draw. I know many children that would draw three fingered humans. Hack some of them can't even draw a single finger. They just have a stick representing the arm.

@@hotrodhunk7389 then you fundamentally don't understand how human brains work. Kids know what hands are, they just either don't *want* to draw it that way or don't think they can. AI *literally* doesn't know what a hand is. It's just stringing together data points as it goes. Oh, also AI steals. It takes art without permission to turn into training data.

Source?

cuz I'm not gonna lie, that sounds like some bullshit my man, gotta prove that assertion. No offense intended.

@vaidenkelsier7757 this is the kind of lie made by someone desperately trying to justify their bad faith arguments and actions. If someone is not interested in learning how Machine Learning works, or doesn't think copying the artwork and styles of others is immoral, they're not going to be interested in arguing in good faith.

@vaidenkelsier7757 I can't drop links because youtube hates links, but if you simply Google the leak list, you'll see that the overwhelming opinion is that the leak isn't a list telling the ai what to look for, but a list made up by people trying to figure out what was scraped. The list is literally just a list of artists the people on reddit think were scraped, with no proof that their artwork was scraped, and more importantly no proof that the AI was told to look for those people in specific. As I said, the list was literally made by people on reddit, and the threads are still up.

@NoESanity the list of artists scraped was released as part of an ongoing court case, which can still be seen, and was only leaked via Reddit. Using the word "literally" in your sentence doesn't make it true. You can "literally" verify this for yourselves.

No he won’t, because it’s not theft and it’s not a crime. None of his work exists in the AI. It’s a bit like taking every recipe from every cooking book and distilling it down to a set of rules as to what a cupcake is, what a muffin is, etc. A cupcake is a muffin with sweeter batter, a lot of muffins contain fruit, etc. The original recipes do not exist in this set of rules, and there is no theft. Anymore than you steal from an author every time you read a book and remember part of it. The original doesn’t exist in the AI, therefore it’s not theft, therefore it’s not a crime

​@@halgariWrong

Its already not monitizable. There was already a court case for that very thing and they were ruled against because it doesn't fall under fair use.

@@asksalottle220 I didn't know that. That does make me happy. Yet if there's not supposed to be money in it, why does everyone go crazy over being able to use AI?

Can't wait until we're calling engineers pathetic for disliking that an AI developer stole everything a human has ever designed and then generates profit for people who have no idea how to engineer anything. Anyway, what do you do?

it defiantly trains on vastly more than that. That is likely the hidden list of "styles" that mid-journey is capable of reproducing if prompted.

Oh the hypocrisy.

Maybe stick to coding and stop stealing people's art :)

@@genericname2747 i do both and use both to make each other better.

@EaterGreen Thanks for admitting you steal art