Art AIs are not competent. They are just a collage generator with a blend tool. They can’t create anything truly new, only an amalgam of pictures they’ve been trained on. They steal art from real artists to do this training. There is no learning process here, the only improvement they can have is through stealing more and more art and getting better at replicating hybrids of various pieces. If they “revolutionise” art related industries, all the art will just be derivative of existing art, there’ll be no soul, no deeper meaning, it will all be empty and shallow.
It’s interesting you mention video games, because I played a game recently that used AI art for character portraits, and they were all lacking any sort of charm, it just vaguely felt like a mugshot image, there was no personality displayed, nothing about their character, just a picture of a person with nothing behind the eyes. The game was a rip-off of another game that literally used characters with circles for heads and dots for eyes and those characters were far more compelling and likeable despite looking technically less proficient. It was an art style with charm, something that art AI can’t actually emulate, it can only try to imitate.
You’re speaking in term of “soul”, I am speaking in term of industry. Industry mass produces. Industry seeks efficiency. Industry seeks profits, not “souls”.
By “Art-related industry”, I am not referring to the few commissions on Pixiv, Twitter, and the likes. These independent artists are not industry.
Ah I see. Yeah, I think we’re in agreement then, it’s much cheaper to have an AI generate a “good enough” bit of media for people to consume than it is to hire artists to actually make something worthwhile. The problem is, those independent artists become industry artists, no one makes furry porn commissions for a living because it’s their lifelong dream. So with fewer artists making art, there’s less art for the AIs to steal from, and they will start to imitate from each other instead of real art. We’ll get the art equivalent of running a sentence through google translate two dozen times and the result will be incomprehensible gibberish. I don’t know if the techbros making this happen actually have a solution for this problem, as it is a societal, not technological one, they can’t just solve it by fine tuning the algorithm. I guess if this AI future comes to pass, except to see every single bit of mass media feeling exactly the same as every other one, just with a vague sci-fi backdrop replaced with a vague fantasy one depending on what trends are projected to be the most profitable.
Now that I think about it, it probably won’t be all that different from how the AAA game industry and Hollywood do things already. Sucks that it means that “furry porn commissions” is the furthest my career can go now though.
But why not just save a step on that first point and just hire the artists to make the art?
I don’t think this will make the current method “obsolete” so much as it will replace it with a cheaper, less meaningful method, one that can mass produce art the same way industrial machinery can produce chairs instead of master artisans making them by hand, but art isn’t something that benefits from that. Companies can benefit from this in the short term, but I can’t imagine people really loving AI generated content as it would all be empty and derivative, even now people are turning away from a lot of traditional entertainment because it is so formulaic and AI art will just compound this issue, it might be cheaper for companies to make, but that doesn’t mean people will buy it in the same quantities as existing media.
But why not just save a step on that first point and just hire the artists to make the art?
Imagine you are an anime producer, do you want to hire an entire studio of illustrators to draw every single frame to produce some 360 hours of 24fps animated video, or do you want to hire a smaller team of artists, make them draw a few key frames, construct an outline, and use an AI to mass produce 1200 hours of 60fps fluid animation? (Of course, the current tech is not there yet, but it is theoretically possible)
For your second paragraph, I feel like you are defining “art” as “art for the sake of art”. But in reality, arts always have a purpose. If not for entertainment, then for sending a message. Arts for the purpose of profit alone would indeed be hollow and soulless (like NFTs for example). Arts that entertain are less hollow, but their value is fast diminishing (like most mainstream entertainment). The most impactful arts are ones that send messages, they will be studied by academic and scholars for centuries to come for their historical relevancy. Regarding AI generated arts, if it has a purpose, it is art. As you can see, hollow art need not to be AI generated, and AI generated need not be hollow.
I suppose you’re right on that first paragraph there, but again, in this hypothetical future, why would anyone spend the time learning to make art when an AI could produce things far more technically proficient than anything they could while learning? If art is truly a dying field, then I would imagine that artists capable of designing work for the AI to play with would be in short supply.
I’m not at all sure I agree with your final paragraph, “art for art’s sake” is an empty phrase, art is about purpose, meaning, humanity, and AI art cannot have purpose, because AI has no motive to make art, it just makes what it is told to do. Even an artist doing furry porn commissions adds their own unique flair to it, that AI art cannot do. To repeat myself, it can imitate, but it cannot emulate. It cannot understand why something makes art good, and the tech guys making these AI art programs are not artists themselves, so they can make something that looks technically proficient, but has no greater meaning behind it. It is hollow, it will always be hollow, as it has none of the humanity that art contains. At best it can have “meaning” in the same way any big checklist based blockbuster film does, something technically proficient, but not something that sticks with the audience, that allows them to reflect or ponder, just big noise. It’s like pop music, it’s getting increasingly “designed” and “constructed” instead of “created” and the result is something technically proficient, that gets stuck in people’s heads, and is often quite pleasant to listen to, but has no deeper substance, it’s all surface level. People do like it, and plenty of pop stars have huge fan followings who adore their work, even if it isn’t “their” work so much as it was songs designed by committee.
But I don’t think everyone is only interested in the shallow surface level versions of media, I do believe people want more. To go back to the furry porn thing again, I have had people genuinely touched by some of the work I’ve created, even in that field, I’ve helped people in a way that AI art cannot do, AI art cannot inspire, it cannot provide meaning, it cannot even be open to interpretation like real art, as the “interpretation” is never “what did the artist mean by this?” but always “what prompts did they use to get that result?” Why should anyone care about an AI art piece someone else “made” when they could just copy paste the prompt and “create” it themselves? Perhaps this could lead to an unexpected result, where if the tech becomes easy enough to use, regular people can create their own animes, entirely without studios, entirely generated by AI, and if that were the case, the studios would struggle to compete, unless they could somehow produce work that is higher quality than what personal use AIs can generate, which would probably loop back around to needing a professional artist’s touch once again.
EDIT: I’m finding this to be a very interesting discussion, I’m not trying to argue with you, I’m just quite enjoying talking about this with you, so I’m hoping I’m not coming across as too abrasive)
in this hypothetical future, why would anyone spend the time learning to make art when an AI could produce things far more technically proficient than anything they could while learning?
Some AI technologies will require precision. You will not prompt precision in human language. You will either has to describe them in a precise domain-specific language or just draw them. For the hypothetical “anime producing AI” tech that I describe, it requires humans to draw key frames and outlines. And if there’s demand, there will be supply.
because AI has no motive to make art, it just makes what it is told to do
AI doesn’t make art by itself, it was a human who told it to do so, and its purpose depends on the human.
The core argument of your second paragraph doesn’t exactly disagree with me, just a different perspective on the same truth: The purpose of an art depends on the intent and purpose of the maker. Most makers will indeed make arts for nothing but a quick buck, which is what already happened in mainstream entertainment even without AI. I don’t yet have any example of meaningful AI arts, but I will never say never.
Perhaps this could lead to an unexpected result, where if the tech becomes easy enough to use, regular people can create their own animes, entirely without studios, entirely generated by AI
Do you consider memes “arts”? I do, as long as these memes convey a message, which they usually do. In this scenario, we will just have more visually impressive memes.
and if that were the case, the studios would struggle to compete, unless they could somehow produce work that is higher quality than what personal use AIs can generate, which would probably loop back around to needing a professional artist’s touch once again.
This is what I meant by “current method will be obsolete”. The studios will need to create even more impressive works which would demand professionals.
Perhaps this could lead to an unexpected result, where if the tech becomes easy enough to use, regular people can create their own animes, entirely without studios, entirely generated by AI
There are already people making animation, or games for that matter, entirely by themselves. AI would allow more people to this easier, although if you are good at making these things without AI it will be easier to get to the desired goal than someone just asking an AI to make a “good shounen anime” and not saying anything more.
Art AIs are actually competent and shall revolutionize Art-related industries (animation, video games, etc).
Text generators are so far not good enough. But I think it still makes some impacts, mostly for the worse:
Art AIs are not competent. They are just a collage generator with a blend tool. They can’t create anything truly new, only an amalgam of pictures they’ve been trained on. They steal art from real artists to do this training. There is no learning process here, the only improvement they can have is through stealing more and more art and getting better at replicating hybrids of various pieces. If they “revolutionise” art related industries, all the art will just be derivative of existing art, there’ll be no soul, no deeper meaning, it will all be empty and shallow.
It’s interesting you mention video games, because I played a game recently that used AI art for character portraits, and they were all lacking any sort of charm, it just vaguely felt like a mugshot image, there was no personality displayed, nothing about their character, just a picture of a person with nothing behind the eyes. The game was a rip-off of another game that literally used characters with circles for heads and dots for eyes and those characters were far more compelling and likeable despite looking technically less proficient. It was an art style with charm, something that art AI can’t actually emulate, it can only try to imitate.
You’re speaking in term of “soul”, I am speaking in term of industry. Industry mass produces. Industry seeks efficiency. Industry seeks profits, not “souls”.
By “Art-related industry”, I am not referring to the few commissions on Pixiv, Twitter, and the likes. These independent artists are not industry.
Ah I see. Yeah, I think we’re in agreement then, it’s much cheaper to have an AI generate a “good enough” bit of media for people to consume than it is to hire artists to actually make something worthwhile. The problem is, those independent artists become industry artists, no one makes furry porn commissions for a living because it’s their lifelong dream. So with fewer artists making art, there’s less art for the AIs to steal from, and they will start to imitate from each other instead of real art. We’ll get the art equivalent of running a sentence through google translate two dozen times and the result will be incomprehensible gibberish. I don’t know if the techbros making this happen actually have a solution for this problem, as it is a societal, not technological one, they can’t just solve it by fine tuning the algorithm. I guess if this AI future comes to pass, except to see every single bit of mass media feeling exactly the same as every other one, just with a vague sci-fi backdrop replaced with a vague fantasy one depending on what trends are projected to be the most profitable.
Now that I think about it, it probably won’t be all that different from how the AAA game industry and Hollywood do things already. Sucks that it means that “furry porn commissions” is the furthest my career can go now though.
There’s nothing that prevents the industry from hiring traditional artists to invent a new art style and feed them to the trainers.
Art AIs can never replace the logic and soul of humans, but they will certainly help copy pasting them easier.
Art AIs don’t make artists obsolete. They will however makes the current method obsolete.
But why not just save a step on that first point and just hire the artists to make the art?
I don’t think this will make the current method “obsolete” so much as it will replace it with a cheaper, less meaningful method, one that can mass produce art the same way industrial machinery can produce chairs instead of master artisans making them by hand, but art isn’t something that benefits from that. Companies can benefit from this in the short term, but I can’t imagine people really loving AI generated content as it would all be empty and derivative, even now people are turning away from a lot of traditional entertainment because it is so formulaic and AI art will just compound this issue, it might be cheaper for companies to make, but that doesn’t mean people will buy it in the same quantities as existing media.
Imagine you are an anime producer, do you want to hire an entire studio of illustrators to draw every single frame to produce some 360 hours of 24fps animated video, or do you want to hire a smaller team of artists, make them draw a few key frames, construct an outline, and use an AI to mass produce 1200 hours of 60fps fluid animation? (Of course, the current tech is not there yet, but it is theoretically possible)
For your second paragraph, I feel like you are defining “art” as “art for the sake of art”. But in reality, arts always have a purpose. If not for entertainment, then for sending a message. Arts for the purpose of profit alone would indeed be hollow and soulless (like NFTs for example). Arts that entertain are less hollow, but their value is fast diminishing (like most mainstream entertainment). The most impactful arts are ones that send messages, they will be studied by academic and scholars for centuries to come for their historical relevancy. Regarding AI generated arts, if it has a purpose, it is art. As you can see, hollow art need not to be AI generated, and AI generated need not be hollow.
I suppose you’re right on that first paragraph there, but again, in this hypothetical future, why would anyone spend the time learning to make art when an AI could produce things far more technically proficient than anything they could while learning? If art is truly a dying field, then I would imagine that artists capable of designing work for the AI to play with would be in short supply.
I’m not at all sure I agree with your final paragraph, “art for art’s sake” is an empty phrase, art is about purpose, meaning, humanity, and AI art cannot have purpose, because AI has no motive to make art, it just makes what it is told to do. Even an artist doing furry porn commissions adds their own unique flair to it, that AI art cannot do. To repeat myself, it can imitate, but it cannot emulate. It cannot understand why something makes art good, and the tech guys making these AI art programs are not artists themselves, so they can make something that looks technically proficient, but has no greater meaning behind it. It is hollow, it will always be hollow, as it has none of the humanity that art contains. At best it can have “meaning” in the same way any big checklist based blockbuster film does, something technically proficient, but not something that sticks with the audience, that allows them to reflect or ponder, just big noise. It’s like pop music, it’s getting increasingly “designed” and “constructed” instead of “created” and the result is something technically proficient, that gets stuck in people’s heads, and is often quite pleasant to listen to, but has no deeper substance, it’s all surface level. People do like it, and plenty of pop stars have huge fan followings who adore their work, even if it isn’t “their” work so much as it was songs designed by committee.
But I don’t think everyone is only interested in the shallow surface level versions of media, I do believe people want more. To go back to the furry porn thing again, I have had people genuinely touched by some of the work I’ve created, even in that field, I’ve helped people in a way that AI art cannot do, AI art cannot inspire, it cannot provide meaning, it cannot even be open to interpretation like real art, as the “interpretation” is never “what did the artist mean by this?” but always “what prompts did they use to get that result?” Why should anyone care about an AI art piece someone else “made” when they could just copy paste the prompt and “create” it themselves? Perhaps this could lead to an unexpected result, where if the tech becomes easy enough to use, regular people can create their own animes, entirely without studios, entirely generated by AI, and if that were the case, the studios would struggle to compete, unless they could somehow produce work that is higher quality than what personal use AIs can generate, which would probably loop back around to needing a professional artist’s touch once again.
EDIT: I’m finding this to be a very interesting discussion, I’m not trying to argue with you, I’m just quite enjoying talking about this with you, so I’m hoping I’m not coming across as too abrasive)
Some AI technologies will require precision. You will not prompt precision in human language. You will either has to describe them in a precise domain-specific language or just draw them. For the hypothetical “anime producing AI” tech that I describe, it requires humans to draw key frames and outlines. And if there’s demand, there will be supply.
AI doesn’t make art by itself, it was a human who told it to do so, and its purpose depends on the human.
The core argument of your second paragraph doesn’t exactly disagree with me, just a different perspective on the same truth: The purpose of an art depends on the intent and purpose of the maker. Most makers will indeed make arts for nothing but a quick buck, which is what already happened in mainstream entertainment even without AI. I don’t yet have any example of meaningful AI arts, but I will never say never.
Do you consider memes “arts”? I do, as long as these memes convey a message, which they usually do. In this scenario, we will just have more visually impressive memes.
This is what I meant by “current method will be obsolete”. The studios will need to create even more impressive works which would demand professionals.
There are already people making animation, or games for that matter, entirely by themselves. AI would allow more people to this easier, although if you are good at making these things without AI it will be easier to get to the desired goal than someone just asking an AI to make a “good shounen anime” and not saying anything more.