The difference is human labor being done by the artist who is creating the art. Ai images are generated using someone elses nonconsensually obtained and uncompensated labor. The prompter is not the artist, the millions of artists used to train ai are.
The nonconsensual acquisition and uncompensated labor are irrelevant. You can train an image generator on entirely paid-for and consensually obtained works, and the output remains not art because it’s produced by a machine incapable of imparting meaning on the images it generates
the output remains not art because it’s produced by a machine incapable of imparting meaning on the images it generates
What is “meaning”, and how would I measure it to determine if an image was created with or without it? Can a human create art without meaning, or is the addition of meaning intrinsic to human actions?
Can a photograph count as art? What’s the difference between a photograph, created by a simple machine which bends light onto a chemical strip, and AI images, which feed letters into a complex equation and tweak colors in a grid to minimize an error equation?
They’re both deterministic processes whereby a human selected input is transformed via a human configured machine into an output.
The difference to me is the degree of effort the human involved puts in. One involves selecting all the parameters and doing composition, where the other is a linguistic composition. The aesthetics or artistry put into the language input is lost because it’s not represented in the output: a low effort throw away prompt can look similar to a very deliberately crafted one. The crafting just provides specificity of output. A skillful and artistic use of words has better and more pleasing modes of expression. Doesn’t mean it can’t happen, just means that it usually doesn’t.
Given that we’re literally in a post about a piece of art that was used to say “fuck your gatekeeping of what constitutes art”, the relish with which people are willing to say “nope, that can’t be art because it’s not traditional” is… Astounding. Right up there with saying “that can’t be art, it doesn’t look like anything”.
It doesn’t have to be measurable. Art is a form of communication, and communication must have meaning behind it. You can’t communicate no idea. Even if you tried, you would be communicating an idea. So no, a human cannot create art without meaning.
Of course a photograph counts as art. A camera is a tool that does exactly what the photographer makes it do. There is direct correlation between the actions of the photographer and the image the camera produces. Positioning, focus, field of view, timing, I’m sure there’s a million other things a photographer could list that act as inputs.
Contrast that with an AI, where you simply describe the picture that you want to see, and it generates a picture based on what you said. You can spend hours, days, weeks perfecting The Prompt so that it creates exactly the image you want to see, and you will never be the one who made the picture, because the machine is the one that made the picture. See paragraph one for why that picture can’t be art.
I never said AI generated pictures aren’t art because they’re not traditional, I said they’re not art because machines can’t think or feel.
How do you mean it doesn’t have to be measurable? You’re actively talking about it’s absence or presence, so how do you know it’s there if you can’t measure it?
I’m also a bit baffled by the assertion that you can’t communicate “no idea”, along with the assertion that an AI generated image doesn’t have meaning.
If it has no meaning, doesn’t that mean it’s communicating “no idea”?
How do you know something has meaning or not if it can’t be measured?
Personally, I think you can only know if a person tells you that they think it has meaning, and that that’s independent of how they made the thing, but I’m curious what you think.
I really don’t see the difference in your camera argument.
Contrast that with a camera, where you simply point it at what you want to see, and it takes a picture. You can spend hours, days, weeks perfecting The Parameters so that it takes exactly the picture you want to see, and you will never be the one who made the picture, because the machine is the one that made the picture. See paragraph one for why that picture can’t be art.
You don’t explain why a machine you control making an image is art in one case and not in the other. I’ve seen where you argue direct correlation, but the prompt is directly correlated to the output, allowing the individual to tweak and change the output. They don’t have total control over the output, but neither does an oil painter, someone blowing glass, or Pollock swinging a brush to create paint splatter. A medium, tool or technique can have limitations.
And what if I’m not telling it what I want? What if I give it a long string of numbers that I’m tweaking until the output matches my wish? That feels a lot less like “describing a commission”, even though it’s the same process.
Personally, I think it’s obvious that AI art is art in the same way that a photograph or using Photoshop can be art. It’s a tool just like any other. It’s just currently more likely to be boring because it invites shallow art, and for what it needs as the artists input it’s more direct to just use that as the art. If you can jam your vision into a prompt, you can almost certainly convey it better with the words themselves, so you’d skip the tool and just write.
Incredible. This tool, unique among all human creations, is immune to human intent. Paint carries it when splashed, dribbled, or exploded with a grenade, but! AI cannot possibly convey a person’s thoughts. The only images that cannot mean anything. No concepts communicated, no commentary made, and obviously nothing to critique.
Yeah, because you aren’t describing the picture you want to see to the brush, and letting it generate the picture for you without any further input. A paint grenade requires a hand to throw it, and there is a direct correlation between what the hand does and what the paint grenade does. There is no such direct correlation when you tell a machine “make an art piece depicting three children in a trench coat trying to sneak into a building labelled ‘the internet’ with a cardboard cutout of a bouncer in their way.” You are a commissioner, not an artist, and the machine isn’t an artist either, which means the image produced is not art.
As if tweaking a prompt seventeen times has less intent than whether your finger twitched while lobbing a paint-covered grenade.
So I’m an art commissioner, except the fictional scenario depicted visually is not art, because it was created by a robot, except the robot can’t create things, so it’s not an artist, because what it makes isn’t art, because it’s not an artist, and oh no I’ve gone cross-eyed.
Again, the prompt that you spend all that time tweaking can be considered literature, which is art. The image you get when you feed that literature into a machine and tell it to do its thing isn’t art.
I compared you to a commissioner because that’s literally the role you’re playing. I didn’t say the robot can’t create things, I said the robot can’t create art. There is no meaning behind the carbdboard cutout of a bouncer because the machine does not know what cardboard or bouncers are. It doesn’t have any thoughts or feelings or opinions or intent regarding anything in the image it produces. It doesn’t know what age verification is, let alone have anything to say about the subject.
You’re being so incredibly disingenuous here. I haven’t made a single circular argument. My position has consistently been that these machines cannot create art because they cannot think or feel. They are not analogous to a brush, because there is no direct correlation between the input you provide and the output it produces.
It doesn’t know what age verification is, let alone have anything to say about the subject.
Yet the image conveyed that it’s about age verification.
You know what the image is about, and if an image is about anything, that is meaning. Should your philosophy require blaming a person, there’s only one human in the loop.
The robot doesn’t know what bouncers or cardboard are, yet it can represent them, on command. Someone used those concepts to visually communicate an abstract thought. You’ve demonstrably understood their intent. Yet you blame the robot, then say robots can’t be blamed, so this image cannot do what it’s already done.
The model is simultaneously too much of an inert tool to ever imbue qualia, and too active a participant to credit any meatbag.
But if I dip a grenade in a paintbucket, then every charred flake goes precisely were I intended.
I’m starting to get bored with this, and I feel like my entire opinion can be summed up by
If I tell a human to draw a cool red dragon sitting on a pile of gold, then I didn’t draw that dragon, that other person did. Replacing that human with an AI doesn’t suddenly mean that I drew it. Ergo, prompt engineers are not visual artists.
Art necessitates intent on the part of an artist, and machines have no intent. The machine has no opinion about why the bouncer is there, nor the implications of children accessing the internet without actual age verification. Those elements are there on the picture solely because that is what was described in the text, and not to convey any meaning.
It is meaningless because the “artist” didn’t mean anything when they made it.
It’s also just not a very interesting piece in general. It’s shallow at best. Just because I can see that the different elements represent different things doesn’t mean it means anything. It doesn’t have anything to say about its very subject matter. They’re just there.
AI isn’t a tool, you do not wield it, you just submit requests and get what you get. You might as well call yourself a chef for placing an order at a restaurant.
I feel that argument struggles in the context of this post though, where the labor is adding a signature onto a toilet. A similar amount of labor went into people typing up their prompts as went in from someone that’s well known adding on a signature. Now, I say this as someone that thinks AI art is wasteful since it uses up so much water and electricity, and is mostly done unethically since the art sourced was mostly done without consent or fair compensation.
Actually, the post’s follow up images posit that he MADE those toilets, since no two are the same and they don’t match existing designs. Saying he put in as little effort as AI “artists” is misunderstanding his work.
That’s the thing, people didn’t care about it till it had a well known person’s name actually attached to the piece though. It was submitted as just a toilet with a signature from an unknown person, it wasn’t originally known that it was from Duchamp.
To me, it feels like there are a few important questions Duchamp poses:
Does art have worth on its own without the artist? Without Duchamp, the piece may well have only been recognized as a toilet with a signature.
Is the artist an integral part to the meaning? I feel like in this case the answer is yes, because it was recognized for more than it was perceived because it became known that it was from Duchamp.
What is the difference between a tool and a piece of art? Namely, where do we draw the line?
Most people think he bought an off-the-shelf urinal, and they still understand, that would be art. It would still function as commentary. The object’s existence in a museum is a jab.
The flipside is-- y’know that handwritten card that goes around? Like ‘I would rather see your scribbliest stick figure than your fanciest AI slop.’ The anti-AI crowd loves it, and it means a lot to them. What would change if we found out it was made in Qwen?
Yes. It’s an added layer of meaning where he put in the effort, and still made a toilet. This is not a low-effort shitpost, even if it looks to be one from a basic perspective. And when you challenge the meaning of art, it says a lot that the effortmade is impossible to determine at a quick glance.
And I don’t know if it means a lot to people, or if it’s just a thing a lot of people agree with. And if it was made in a shitty ai slop generator, then it would change it from “defiant speach” to “I don’t agree with what I say, I just want you to like me.” And based on every piece of AI I have ever seen, even the stuff AI bros like to boast about, there would be a spelling error they didn’t notice.
The low-effort shitpost everyone thinks it is, would still have meaning.
“I don’t agree with what I say, I just want you to like me.”
I will remind you we’re talking about Fountain. Duchamp trolled the art world by meticulously recreating something trivial. If someone popularized an anti-AI screed, using AI, that would be funniest goddamn thing - and a complete rebuttal.
Imagine the glee in that reveal. Some guy spent hours rendering, tweaking, rendering, tweaking, giggling to himself the whole time, until it produced exactly the image you’ve seen. All flaws excused by the proud AI haters, the same way they (and its text) would excuse literally anything a hand does with a pen. And then - ta-da! Here’s the prompt and the seed! You all found meaning in a generated image, and experienced emotional connection because of it. Get loved, idiot. Get camaraderie’d.
Then you get to watch people twist in knots. It never meant anything to them! Anymore! It’s complete trash, and they’re throwing out all their favorite band’s t-shirts, because they always hated them. It tricked them… but not in any way that confers intent.
Jesus christ, you’re having more delusions than AI. Do you honestly think anyone would care about a picture of some text they agree with so much that the rug pull will affect them in any meaningful way? Do you think that guy hitting refresh on his prompt generator is putting in meaningful effort? Do you think anyone but AI chuds will excuse the blatantly obvious flaws the way you guys do?
Even at this second, I can’t even remember what the image looks like. It doesn’t mean anything to me. I just agree with the words.
I’m sure your little fiction brings you comfort, but it’s just fiction.
It doesn’t mean anything to me. I just agree with the words.
… the words are its meaning.
Those words say: flaws are fine, for anything done by hand.
If Fountain was an off-the-shelf urinal, would you go crack it with a hammer? That’s how people genuinely talk about generated images. They latch onto terms like intent and declare the robot cannot provide it, therefore, some image they quite enjoyed is now a useless husk. As if they can delete the emotional response from their brain.
Like a clever visual metaphor doesn’t parse, if it was rendered instead of scribbled.
Ah, so it wasn’t art for the years that people thought it was a store bought urinal? And it currently exists in a superposition state of art and not-art because it’s not actually known if he sculpted it or not?
I certainly agree that intellectual property is bullshit and should be abolished with the caveat that it should be abolished after capitalism is done away with. However, labor value ≠ intellectual property. I don’t care that the stolen labor used in the end product is minimal, I care that the labor was done by artists and used without compensation by a corporate entity. I care that someone then takes advantage of those artists labor by prompting a machine who uses that stolen labor to generate their idea, for which they had the most minimal intentional input, and calls themselves an artist. I also care that throughout this entire process a corporation is making money off a process that would be impossible without the artists. You say culture belongs to us all but ai cannot produce culture, it offers nothing new, it is an aggregate of ideas that already exist. Cultures change a grow, they are living things that ai offers nothing to.
Tools don’t have culture, people do. New subcultures exist for this tool. Like primitive CGI, it produces images differently. Onlookers seeing the limitations, and offering “Well why don’t you just draw that?,” do not get it.
Would you feel any differently if some collective of artists created a model entirely from their own works, specifically so people can describe images into existence? The end result would not be any different.
I would feel differently about that actually. I still wouldn’t call anything created through prompt a piece of art created by the prompter though. I might call the ai itself a form of art. I might call the prompted images art but the art would have come from and belong to the artists who created the ai
Why? They didn’t make those images, any more than you would. They made the thing that made those images.
The artists did make the images generated by their ai. They put their labor into creating the art that feeds it and they have the ability to curate what art goes into the ai in order to give it a meaning or message. The major factors in what I consider art are labor, intent, and meaning. The prompter lacks these. Their labor, if you would consider prompting to be labor, is absolutely minimal, you could not possible input less labor. Their intent is missing. Think of an oil painting for example. Each brush stroke is intentional, they all have purpose whether it be for representing the beauty of a scene through the lense that they see it and want the viewer to see it or for creating a message they want the viewer to understand. How can the prompter accomplish this? The prompter might give the image meaning but how is the observer meant to understand it when the prompter cannot guide them?
The ai these artists built and curated might be able to translate their message, guided by their intent, and created by their labor if built for that purpose and for that reason I consider images created by it to be their art. Who prompts the ai makes no difference.
I include might, in both my comments because this remains to be seen.
Sorry, taking this seriously. So… if I use their tool… and enter a lurid combination of popular cartoon characters, morally indefensible fetishes, and prominent political figures… it’s their fault. They made that art. The image you closed immediately but cannot unsee, contains absolutely none of my meaning. It is immaculate of my intent, and I cannot be blamed.
No matter how many hours I spent getting Frieren’s spit to land in Starmer’s ear.
If they created the ai to be capable of producing that sort of image then yes. They curated the ai, they gave it the data that allowed it to produce that. You can’t train an ai on, as a random example, exclusively traditonal oil paints pre 2000 and have it make porn of frieren. They would have to intentionally give it that capability. Should they have given it that ability, doesn’t it say something about their intent and meaning in constructing the ai? Isn’t it purposeful on their part?
It doesn’t retain pixels at all. A better way to describe what it retains (though not accurate) is brush strokes. It retains much more of the information than the raw byte count could imply. It’s effectively compressed by capturing relationships of pixels rather than the pixels themselves.
You and I know that - critics might not. They talk like image files go into the model and stay there. One guy insisted that training was no different from encoding a JPEG, because… numbers.
The information gleaned from any single image is hilariously minimal. It’s insulting that anything recognizable comes out. A prolific artist’s contribution means their style is distinguished from the rest of humanity by only a handful of bits.
The difference is human labor being done by the artist who is creating the art. Ai images are generated using someone elses nonconsensually obtained and uncompensated labor. The prompter is not the artist, the millions of artists used to train ai are.
The nonconsensual acquisition and uncompensated labor are irrelevant. You can train an image generator on entirely paid-for and consensually obtained works, and the output remains not art because it’s produced by a machine incapable of imparting meaning on the images it generates
What is “meaning”, and how would I measure it to determine if an image was created with or without it? Can a human create art without meaning, or is the addition of meaning intrinsic to human actions?
Can a photograph count as art? What’s the difference between a photograph, created by a simple machine which bends light onto a chemical strip, and AI images, which feed letters into a complex equation and tweak colors in a grid to minimize an error equation?
They’re both deterministic processes whereby a human selected input is transformed via a human configured machine into an output.
The difference to me is the degree of effort the human involved puts in. One involves selecting all the parameters and doing composition, where the other is a linguistic composition. The aesthetics or artistry put into the language input is lost because it’s not represented in the output: a low effort throw away prompt can look similar to a very deliberately crafted one. The crafting just provides specificity of output. A skillful and artistic use of words has better and more pleasing modes of expression. Doesn’t mean it can’t happen, just means that it usually doesn’t.
Given that we’re literally in a post about a piece of art that was used to say “fuck your gatekeeping of what constitutes art”, the relish with which people are willing to say “nope, that can’t be art because it’s not traditional” is… Astounding. Right up there with saying “that can’t be art, it doesn’t look like anything”.
It doesn’t have to be measurable. Art is a form of communication, and communication must have meaning behind it. You can’t communicate no idea. Even if you tried, you would be communicating an idea. So no, a human cannot create art without meaning.
Of course a photograph counts as art. A camera is a tool that does exactly what the photographer makes it do. There is direct correlation between the actions of the photographer and the image the camera produces. Positioning, focus, field of view, timing, I’m sure there’s a million other things a photographer could list that act as inputs.
Contrast that with an AI, where you simply describe the picture that you want to see, and it generates a picture based on what you said. You can spend hours, days, weeks perfecting The Prompt so that it creates exactly the image you want to see, and you will never be the one who made the picture, because the machine is the one that made the picture. See paragraph one for why that picture can’t be art.
I never said AI generated pictures aren’t art because they’re not traditional, I said they’re not art because machines can’t think or feel.
How do you mean it doesn’t have to be measurable? You’re actively talking about it’s absence or presence, so how do you know it’s there if you can’t measure it?
I’m also a bit baffled by the assertion that you can’t communicate “no idea”, along with the assertion that an AI generated image doesn’t have meaning.
If it has no meaning, doesn’t that mean it’s communicating “no idea”?
How do you know something has meaning or not if it can’t be measured?
Personally, I think you can only know if a person tells you that they think it has meaning, and that that’s independent of how they made the thing, but I’m curious what you think.
I really don’t see the difference in your camera argument.
You don’t explain why a machine you control making an image is art in one case and not in the other. I’ve seen where you argue direct correlation, but the prompt is directly correlated to the output, allowing the individual to tweak and change the output. They don’t have total control over the output, but neither does an oil painter, someone blowing glass, or Pollock swinging a brush to create paint splatter. A medium, tool or technique can have limitations.
And what if I’m not telling it what I want? What if I give it a long string of numbers that I’m tweaking until the output matches my wish? That feels a lot less like “describing a commission”, even though it’s the same process.
Personally, I think it’s obvious that AI art is art in the same way that a photograph or using Photoshop can be art. It’s a tool just like any other. It’s just currently more likely to be boring because it invites shallow art, and for what it needs as the artists input it’s more direct to just use that as the art. If you can jam your vision into a prompt, you can almost certainly convey it better with the words themselves, so you’d skip the tool and just write.
I sorta got into that in my other comment
Incredible. This tool, unique among all human creations, is immune to human intent. Paint carries it when splashed, dribbled, or exploded with a grenade, but! AI cannot possibly convey a person’s thoughts. The only images that cannot mean anything. No concepts communicated, no commentary made, and obviously nothing to critique.
Yeah, because you aren’t describing the picture you want to see to the brush, and letting it generate the picture for you without any further input. A paint grenade requires a hand to throw it, and there is a direct correlation between what the hand does and what the paint grenade does. There is no such direct correlation when you tell a machine “make an art piece depicting three children in a trench coat trying to sneak into a building labelled ‘the internet’ with a cardboard cutout of a bouncer in their way.” You are a commissioner, not an artist, and the machine isn’t an artist either, which means the image produced is not art.
As if tweaking a prompt seventeen times has less intent than whether your finger twitched while lobbing a paint-covered grenade.
So I’m an art commissioner, except the fictional scenario depicted visually is not art, because it was created by a robot, except the robot can’t create things, so it’s not an artist, because what it makes isn’t art, because it’s not an artist, and oh no I’ve gone cross-eyed.
If it’s not art, what are we looking at?
Again, the prompt that you spend all that time tweaking can be considered literature, which is art. The image you get when you feed that literature into a machine and tell it to do its thing isn’t art.
I compared you to a commissioner because that’s literally the role you’re playing. I didn’t say the robot can’t create things, I said the robot can’t create art. There is no meaning behind the carbdboard cutout of a bouncer because the machine does not know what cardboard or bouncers are. It doesn’t have any thoughts or feelings or opinions or intent regarding anything in the image it produces. It doesn’t know what age verification is, let alone have anything to say about the subject.
You’re being so incredibly disingenuous here. I haven’t made a single circular argument. My position has consistently been that these machines cannot create art because they cannot think or feel. They are not analogous to a brush, because there is no direct correlation between the input you provide and the output it produces.
A picture.
Yet the image conveyed that it’s about age verification.
You know what the image is about, and if an image is about anything, that is meaning. Should your philosophy require blaming a person, there’s only one human in the loop.
The robot doesn’t know what bouncers or cardboard are, yet it can represent them, on command. Someone used those concepts to visually communicate an abstract thought. You’ve demonstrably understood their intent. Yet you blame the robot, then say robots can’t be blamed, so this image cannot do what it’s already done.
The model is simultaneously too much of an inert tool to ever imbue qualia, and too active a participant to credit any meatbag.
But if I dip a grenade in a paintbucket, then every charred flake goes precisely were I intended.
I’m starting to get bored with this, and I feel like my entire opinion can be summed up by
If I tell a human to draw a cool red dragon sitting on a pile of gold, then I didn’t draw that dragon, that other person did. Replacing that human with an AI doesn’t suddenly mean that I drew it. Ergo, prompt engineers are not visual artists.
Art necessitates intent on the part of an artist, and machines have no intent. The machine has no opinion about why the bouncer is there, nor the implications of children accessing the internet without actual age verification. Those elements are there on the picture solely because that is what was described in the text, and not to convey any meaning.
It is meaningless because the “artist” didn’t mean anything when they made it.
It’s also just not a very interesting piece in general. It’s shallow at best. Just because I can see that the different elements represent different things doesn’t mean it means anything. It doesn’t have anything to say about its very subject matter. They’re just there.
A flimsy guard against children acting adult on the internet has nothing to say, says meaning understander.
Someone should’ve scribbled it on a napkin, then it would intend to communicate an idea. It would be art!
AI isn’t a tool, you do not wield it, you just submit requests and get what you get. You might as well call yourself a chef for placing an order at a restaurant.
Writing a recipe doesn’t make you a chef, but you still came up with that dish.
It’s still food.
I feel that argument struggles in the context of this post though, where the labor is adding a signature onto a toilet. A similar amount of labor went into people typing up their prompts as went in from someone that’s well known adding on a signature. Now, I say this as someone that thinks AI art is wasteful since it uses up so much water and electricity, and is mostly done unethically since the art sourced was mostly done without consent or fair compensation.
Actually, the post’s follow up images posit that he MADE those toilets, since no two are the same and they don’t match existing designs. Saying he put in as little effort as AI “artists” is misunderstanding his work.
That’s the thing, people didn’t care about it till it had a well known person’s name actually attached to the piece though. It was submitted as just a toilet with a signature from an unknown person, it wasn’t originally known that it was from Duchamp.
To me, it feels like there are a few important questions Duchamp poses:
Serious question: does that matter?
Most people think he bought an off-the-shelf urinal, and they still understand, that would be art. It would still function as commentary. The object’s existence in a museum is a jab.
The flipside is-- y’know that handwritten card that goes around? Like ‘I would rather see your scribbliest stick figure than your fanciest AI slop.’ The anti-AI crowd loves it, and it means a lot to them. What would change if we found out it was made in Qwen?
Yes. It’s an added layer of meaning where he put in the effort, and still made a toilet. This is not a low-effort shitpost, even if it looks to be one from a basic perspective. And when you challenge the meaning of art, it says a lot that the effortmade is impossible to determine at a quick glance.
And I don’t know if it means a lot to people, or if it’s just a thing a lot of people agree with. And if it was made in a shitty ai slop generator, then it would change it from “defiant speach” to “I don’t agree with what I say, I just want you to like me.” And based on every piece of AI I have ever seen, even the stuff AI bros like to boast about, there would be a spelling error they didn’t notice.
The low-effort shitpost everyone thinks it is, would still have meaning.
I will remind you we’re talking about Fountain. Duchamp trolled the art world by meticulously recreating something trivial. If someone popularized an anti-AI screed, using AI, that would be funniest goddamn thing - and a complete rebuttal.
Imagine the glee in that reveal. Some guy spent hours rendering, tweaking, rendering, tweaking, giggling to himself the whole time, until it produced exactly the image you’ve seen. All flaws excused by the proud AI haters, the same way they (and its text) would excuse literally anything a hand does with a pen. And then - ta-da! Here’s the prompt and the seed! You all found meaning in a generated image, and experienced emotional connection because of it. Get loved, idiot. Get camaraderie’d.
Then you get to watch people twist in knots. It never meant anything to them! Anymore! It’s complete trash, and they’re throwing out all their favorite band’s t-shirts, because they always hated them. It tricked them… but not in any way that confers intent.
Yes, it would have meaning, but less.
Jesus christ, you’re having more delusions than AI. Do you honestly think anyone would care about a picture of some text they agree with so much that the rug pull will affect them in any meaningful way? Do you think that guy hitting refresh on his prompt generator is putting in meaningful effort? Do you think anyone but AI chuds will excuse the blatantly obvious flaws the way you guys do?
Even at this second, I can’t even remember what the image looks like. It doesn’t mean anything to me. I just agree with the words.
I’m sure your little fiction brings you comfort, but it’s just fiction.
… the words are its meaning.
Those words say: flaws are fine, for anything done by hand.
If Fountain was an off-the-shelf urinal, would you go crack it with a hammer? That’s how people genuinely talk about generated images. They latch onto terms like intent and declare the robot cannot provide it, therefore, some image they quite enjoyed is now a useless husk. As if they can delete the emotional response from their brain.
Like a clever visual metaphor doesn’t parse, if it was rendered instead of scribbled.
Stupid question, of course it matters
Ah, so it wasn’t art for the years that people thought it was a store bought urinal? And it currently exists in a superposition state of art and not-art because it’s not actually known if he sculpted it or not?
Use your words.
Fuck intellectual property. Culture belongs to us all.
And a 20 GB model trained on a billion images retains less than eight pixels of each.
I certainly agree that intellectual property is bullshit and should be abolished with the caveat that it should be abolished after capitalism is done away with. However, labor value ≠ intellectual property. I don’t care that the stolen labor used in the end product is minimal, I care that the labor was done by artists and used without compensation by a corporate entity. I care that someone then takes advantage of those artists labor by prompting a machine who uses that stolen labor to generate their idea, for which they had the most minimal intentional input, and calls themselves an artist. I also care that throughout this entire process a corporation is making money off a process that would be impossible without the artists. You say culture belongs to us all but ai cannot produce culture, it offers nothing new, it is an aggregate of ideas that already exist. Cultures change a grow, they are living things that ai offers nothing to.
Tools don’t have culture, people do. New subcultures exist for this tool. Like primitive CGI, it produces images differently. Onlookers seeing the limitations, and offering “Well why don’t you just draw that?,” do not get it.
Would you feel any differently if some collective of artists created a model entirely from their own works, specifically so people can describe images into existence? The end result would not be any different.
I would feel differently about that actually. I still wouldn’t call anything created through prompt a piece of art created by the prompter though. I might call the ai itself a form of art. I might call the prompted images art but the art would have come from and belong to the artists who created the ai
Why? They didn’t make those images, any more than you would. They made the thing that made those images.
Attributing authorship if they write the prompt, but not if you write the prompt, is a religious debate.
The artists did make the images generated by their ai. They put their labor into creating the art that feeds it and they have the ability to curate what art goes into the ai in order to give it a meaning or message. The major factors in what I consider art are labor, intent, and meaning. The prompter lacks these. Their labor, if you would consider prompting to be labor, is absolutely minimal, you could not possible input less labor. Their intent is missing. Think of an oil painting for example. Each brush stroke is intentional, they all have purpose whether it be for representing the beauty of a scene through the lense that they see it and want the viewer to see it or for creating a message they want the viewer to understand. How can the prompter accomplish this? The prompter might give the image meaning but how is the observer meant to understand it when the prompter cannot guide them?
The ai these artists built and curated might be able to translate their message, guided by their intent, and created by their labor if built for that purpose and for that reason I consider images created by it to be their art. Who prompts the ai makes no difference.
I include might, in both my comments because this remains to be seen.
But does the JPEG have Buddha nature?
Sorry, taking this seriously. So… if I use their tool… and enter a lurid combination of popular cartoon characters, morally indefensible fetishes, and prominent political figures… it’s their fault. They made that art. The image you closed immediately but cannot unsee, contains absolutely none of my meaning. It is immaculate of my intent, and I cannot be blamed.
No matter how many hours I spent getting Frieren’s spit to land in Starmer’s ear.
If they created the ai to be capable of producing that sort of image then yes. They curated the ai, they gave it the data that allowed it to produce that. You can’t train an ai on, as a random example, exclusively traditonal oil paints pre 2000 and have it make porn of frieren. They would have to intentionally give it that capability. Should they have given it that ability, doesn’t it say something about their intent and meaning in constructing the ai? Isn’t it purposeful on their part?
Lol somebody invested in AI
It doesn’t retain pixels at all. A better way to describe what it retains (though not accurate) is brush strokes. It retains much more of the information than the raw byte count could imply. It’s effectively compressed by capturing relationships of pixels rather than the pixels themselves.
You and I know that - critics might not. They talk like image files go into the model and stay there. One guy insisted that training was no different from encoding a JPEG, because… numbers.
The information gleaned from any single image is hilariously minimal. It’s insulting that anything recognizable comes out. A prolific artist’s contribution means their style is distinguished from the rest of humanity by only a handful of bits.