AI Does Not Make Art

 AI Does Not Make Art

Let’s not bury the lede. Everyone already knows that “AI art” is not art. Even strong advocates of it do not discuss it like art, they don’t treat the output like art, and no one would describe an AI robot as an artist. The point of this essay is to discuss the meaning of art and what makes this art or that not, and why the question of “what is art?” is fundamental to the process of making it. And how all this relates to why we call something which we know not to be art “art”.

Probably the first thing I think of when I hear “art” is statues and paintings, but the term’s usage is actually quite broad. Well, whether videogames can be considered art might depend on whether the reader believes that art must be non-interactive, or art must be created by a singular auteur, or a number of other factors, What about “Modern Art”? Some people will look at its abstract shapes and see art, some will see just mess and chaos.

Or postmodern art. “This is not a pipe.” Makes ya think.

Well, things might not always be criticized for whether or not they are art, but whether they are good, by some definition of “good”. It’s an ongoing conversation. What is good? What is good about art?

The nazis hated modern art, calling it “degenerate” and so they burned it. I feel that the questions of “what is art?” and “what is good?” are somewhat linked. Bad art is still technically art, but it’s not good. But if it’s good, it is “real” art. Art is supposed to instill certain emotions in you and that which does must be art. But if it encourages the viewer to rebel or think critically of the powers that be, then it’s not doing what art is supposed to do. It’s bad. Even today, litigating over what is or is not art, or what is or is not good, can sometimes be linked to fascistic ideations; a topic which has been covered a lot in the past decade, but I liked Jacob Geller’s “Who’s Afraid of Modern Art: Vandalism, Video Games, and Fascism”.

Personally, I think just the question “is this art?” is interesting enough. When I was working on my song, “BWV912 with Pirates,” I started playing around with panning effects and kept thinking… I think they make my instruments sound more evocative and alive, but why do other songs not do it? Are panning effects legitimate forms of self expression, or rather a distraction from the music that could be heard more clearly if the instruments would just stay still? What if what makes it art is simply that I intended it to be this way? What if that itself is pretentious bullshit because objective standards are there for a reason? What if the question of whether or not it is art was the message of my art?

Is it creative for Jackson Pollock to drop paint onto a canvas? I feel like a lot of the answer to that relates to how much agency you assign Pollock to how the painting came out. Is there meaning to each splotch and drop or is it more about the bigger picture? The chaotic mixing of colors yet with a sense of order and intention? Or is it merely the idea that Pollock’s paintings have meaning be what gives them meaning? Is Pollock’s reputation not partially what drives the value of his paintings and not their intrinsic properties?

Does art have meaning, or is meaning something we assign it? If a work of art means something different to two different people, is one understanding of its meaning more “correct” than the other? Does art speak to universal truths, or merely our desire to find meaning? What is art?

It’s 2025; if you want a picture, you can just ask an AI robot to generate it and it’ll look pretty much like how you described. It’ll render your picture in whatever style you like, have anything you like be in it, have any special effects you like, you name it. But is it art? 

“Tsukuyomi: The Divine Hunter”, released earlier this year, promised “A New World from Visionary Creator Kazuma Kaneko” on its Steam store page. It does legitimately contain a lot of his art, but the game intended to be an endless supply of the artist’s work through its “Creation” system. Going through the levels, you’d make certain decisions and those somehow influence an AI to draw you a card. But the cards are kinda ass. And they plagiarize other artists and trademarks. In the game, there’s a voting system where players can look at two generated cards and give feedback to the AI about which is better. 

I think there’s a point to be made about artistic vision. I was not a fan of all of Kaneko-sensei’s character designs, but I always appreciated his distinct styles. Both when I see a character design from Kaneko-sensei that I like and that I dislike, I learn something about myself and the artist’s tastes. So the idea of having players dictate what art is good and what is not runs the risk of the AI and Tsukuyomi community as a whole sliding further and further from Kaneko-sensei’s vision. The art in the game may get better in accordance with the tastes of the community, but they do not have artistic authority to make the decision of what is more or less Kaneko-like. An artist is not a static idea, they can and will choose to change their styles over time, integrate new elements into their work, take elements out, innovate. The best fans can do here is preserve what Kaneko-sensei meant to them in past glory. The AI can experiment on the Kaneko-sensei who worked on the original Persona, but not the one alive today.

AI simulates art, or a specific fantasy of art where the artist imagines “boy by the park” and then proceeds to paint a boy by a park. It does not ask “what is art?” nor does it think about its place in the art world. It does not consider the tradeoffs between objective standards of quality vs self expression. It doesn’t think about how the way it chooses to express this idea will affect how its art is perceived or what it means. It simply produces to the best of its ability what you probably wanted when you wrote the prompt.

With a Jackson Pollock painting, one might be able to question whether what he did was creative or not, intentional or not, art or not, but there is no creativity to AI “art”, nor any intent. It is a machine which serves its master and does not have a will of its own. What it produces reveals very little about it and says even less. AI “art” doesn’t have an inherent meaning. We can’t debate why it was decided that Luke Skywalker in “Skywalker Taking a Bath With Jesus” has six fingers and two hands, not one. “AI struggles with hands,” they’ll say. If Picaso pulled that shit, art critics would be debating why he drew it that way until the end of time, but it would be absurd to assume that any part of an AI-produced image was made with artistic integrity. Any aspect of the drawing can only be discussed in terms of the quality of the model that produced it.

That AI art is not art is something so obvious that even AI art advocates understand this on a fundamental level and do not talk about it as if it were art. So why do we call it “art”? 

I do not really think AI advocates are too stupid to recognize that there is a difference between “AI art” and what artists do. So that leaves us with two possibilities: Either they know there’s a difference but still think both are “art”, or we simply don’t have a better word for what AI produces. “AI pictures” just doesn’t roll off the tongue like “AI art”. But let’s consider the former.

It is not impossible for AI pictures to convey meaning. I could ask ChatGPT to generate an image of “Elon Musk and DOGE with armed law enforcement breaking into Trump's office and declaring that the whitehouse is a drain on taxpayer money” and this image would convey my feelings about the self-destructive and contradictory nature of the Trump administration and DOGE. But I couldn’t say that I made it. The resulting rendition would be, in art terms, the AI’s “vision” of the scene, but it would not be even a pale imitation of my own imagination. Each brush stroke–pretending for a moment that AI actually uses technique–would not be in service of the message I wanted to convey, only the description of the image that I asked for.

Now, I intended to put the generated image in this article, but big surprise, I can’t actually make ChatGPT generate this image.

“I can't create an image with real people (like Elon Musk or Donald Trump) in politically sensitive or potentially misleading scenarios—especially involving law enforcement or controversial statements—because it could spread misinformation or violate our content guidelines.”

So I pressed further:

I like how ChatGPT here thought that “This place is a money pit!” would be a satire on government spending rather than my stated intention of satirizing that attitude. ChatGPT can’t speak for me because it doesn’t understand me. It doesn’t understand anything..

AI pictures are inherently meaningless and I think that’s part of the point. Calling AI pictures “art” degrades art by assuming that art is inherently meaningless. You don’t see a lot of people generating new Picasso-style paintings or experimenting with pointillism or morphism. Advocates of AI pictures tend to want to generate bland, traditional, boring subjects. If something looks really high quality, but doesn’t seem to be saying anything, that’s just the AI aesthetic.

The fascists viewed subversive art as either “bad art” or not art at all. AI is incapable of subversion and can certainly not subvert the expectations of the prompter. Which is kind of perfect, isn’t it? AI art cannot challenge us and it is explicitly restricted from making certain types of challenging images. AI art does not speak truth to power and by inhibiting speech critical of power, works to its advantage. That’s without even considering the ways in which those with power can manipulate AI output and have. 

Anti-intellectualism has had a major victory just by the linguistics of calling AI art “art”. It suggests that the thought, effort, and attention that goes into making art is not a part of what makes it “art,” only the final product matters. Which is funny because the final product is never what most people would call “good”. Advocates of AI never seem to have any real discernment between good and bad art.

I was really interested in writing back in highschool and I remember hearing this quote or something. It said you really don’t get better as an artist as you practice. What improves is your taste. In constant efforts to improve and perfect your craft, you will discover your preferences over time, learn techniques that improve your execution, and what you release to the public will always be a pale imitation of what you wanted to create. You’ll always wish you had more time to work on it or you were just better than you are. But you’ll grow, get smarter and better and over time look back at past projects both for their strengths and their flaws. You also know that the thing you’re working on right now will be the same. You will never be good enough. You will never achieve perfection. And yet you continue to write, paint, design, sing, compose, and play. There is a drive, burning inside of you, screaming to get out. And because words cannot describe it, you turn to abstract representations of your thoughts and feelings. Art. 

I don’t know what art is and I don’t know why we need to create it, but I think the fact that we need to create it is itself fascinating. We need to create art even though we know it will never be good enough. I’d like to teach the world to sing in perfect harmony; I’d like to buy the world a coke but I know that a song cannot enact world peace. Probably everyone who has ever written an anti-war song knew that, and maybe some day an artist will come around whose anti-war is so universally liked that peace does finally win. But it’s probably not going to be you, and probably won’t happen in your lifetime.They wrote those songs despite knowing this. Why do we do this to ourselves? Why are we so driven to try to achieve goals that are impossible? Can art save the world?

When I played Clair Obscure: Expedition 33, I had a lot of issues with the introduction. There were numerous graphical glitches, a super sonic Maelle, pacing issues, and I was disappointed not to have learned more about the society from which we were departing. To some degree, I have to give Sandfall Interactive artistic license about how they wanted to do their game, but after learning about the story of the game’s creation, I just couldn’t bear to fault it for anything. Guy leaves Ubisoft because he wants to work on his own games, finding his work in Corporate unrewarding. Finds people from all over the internet to form his starting staff. Hires a small team of largely beginning developers. Signs with the publisher Kepler Interactive which is itself cooperatively run by developers ensuring artistic freedom and dev-friendly practices. And the result is a game that is engrossing, beautiful, and meaningful.

When I think about how difficult Expedition 33 must have been to make, these subtle flaws vanish away and I am left awestruck by what they were able to achieve. It’s mainstream in its aesthetics which lets it compete with AAA games, but also experimental and artistic. Risky. And the combat is immediately satisfying. I think about the intensity I feel when I have finally memorized a boss’s attacks and have equipped all the right pictos to counter its attacks, then execute those timings perfectly and am rewarded for that effort by getting to see the victory screen. The feeling of personal accomplishment is mixed with an appreciation for how the challenge of developing the game must have been orders of magnitude harder than what I just did. A false rumor spread online that Expedition 33 was made by 30 ex-Ubisoft employees who wanted to fight back against the corporate elites of the gaming industry. They’d look at the number of consecutive players on Steam vs other “AAA” games. If you were playing E33, you were making a political statement about who should make games, how, and why. It is impossible to separate Expedition 33 from the story of its development.

I call it “Man asks for watercolor drawing, but gets drawing with watercolor brushes in it instead.” The story behind this piece is I thought it would be funny to contrast the impossibility of Expedition 33’s existence with how low effort AI art is. It doesn’t take months or years to build up a portfolio of an AI artist, a couple good brainstorming sessions will do. When I embarked on my journey to make the above masterpiece, I did not have to consider whether or not I have any watercolor brushes or the right colors, nor if I have the technical competence to pull off some of the brush stroke techniques I’d want to employ, nor how much time, energy, and physical space on my desk creating the picture would require. If I had done any of those things, I would have wanted to pick a better subject. I just wrote the prompt to Gemini and looked at the first image and said “that’ll do.” 

There is the concept of AI-assisted art where AI and artist work together to refine an image, but every pixel representative of the AI’s output represents an aspect of the image for which no conscious thought, effort, or intent was put into it. Eventually, the artist just has to decide if it’s wrong, or “that’ll do.”

AI images serve a different social purpose from art. Whereas art is created because it, for some reason, needs to exist, AI images are often used for more functional purposes. Games on Steam, in efforts to reduce development costs while meeting consumer demands, have increasingly been using AI to generate store capsules and in-game assets. For the most part, gamers don’t seem to mind, though the assets generated can often be of low quality and this sometimes comes up in reviews. Players are generally aware of when AI is used because it just looks or sounds like AI-generated content, but they don’t ultimately think the game is bad for having it. They know they are playing budget games and by knowing this, the subtle flaws melt away and they are just amazed that this dev who could never afford to hire voice actors has AI digitized voices, or who could never afford an artist has character portraits for all NPCs.

We can generate images as jokes to post to discord, or to try and express something we’re imagining, or to create assets in a video game. It is, above all else, convenient. But it cannot convey to its viewer that which is indescribable. And maybe neither can art. Maybe we need so many forms of artistic expression in part because in all this time we haven’t found a medium that truly allows us to convey the messages our souls are screaming to get out. Art cannot serve any specific purpose or solve real problems because the production of it is too inefficient and it generally doesn’t solve anything, but its lack of efficiency and uselessness is also what instills it with meaning. 

I recently became struck while walking through the store by the bold colors on this box’s design. It’s a visual style that I’d like to use for a project of mine. But is it “art”? Is “commercial art” an oxymoron? Is this one art because I choose to appreciate it as art?

I think it’s at least fair to say there are shades of grey. Every aspect of this box, from the choice of colors to the composition of elements, are intended to simultaneously catch the consumer’s eyes while informing them. Aspects of this design are functional and brutalist, while others I find surprisingly striking. Someone had an idea about how to use these strong blues and reds in a way that it’s not typically done, but conveys the ideas they thought needed conveying. But I wouldn’t hang it on my wall.

Typography is an art and the Coca Cola logo might at one point have had artistic merit, but it’s hard for me to view it as art when for my entire life it has represented a soda rather than being an aesthetically pleasing image.

Things just stop looking like art when they’re made for specific reasons. It’s about the relationship between artist, audience, and the art itself. Asking “What is the artist trying to convey to me through this medium?” is a part of the joy in experiencing art, but an advertisement or propaganda is generally going to be blunt and upfront. There’s no mystery to solve.

As video games and movies become more expensive to produce, studios become more and more risk averse. In indie gaming, even groundbreaking titles like World of Goo, Spelunky, and Slay the Spire are following the familiar pattern of releasing sequels rather than trying new ideas, breaking new grounds. Indie gaming is no longer known for being necessarily more innovative than the AAA industry, just an alternative to its even more risk-averse practices.

Art in its purest form cannot survive capitalism as art is inherently inefficient whereas capitalism optimizes for profit. It is no wonder we don’t know what art is. We’ve been living inside a capitalist society so long, the slop an AI can produce has become indistinguishable.

I am probably not the wisest or most intelligent or most informed person to speak on AI and art. I didn’t write this to flex authority I don’t have, nor because I think I have anything particularly novel to say on the subject. I wrote it because I couldn’t stop myself. I felt the burning drive to create. And then I did it. I sat down and wrote out every word. I thought about whether this line would flow better without a comma or that one should be broken up into two. 

But what’s the point? No one is going to read this and say “wow, I thought AI art was kinda cool, but when you put it that way…” And it’s not like I’m calling for some specific action to be taken that would right what I think is wrong. There was no specific reason to write this, it’s just that knowing the futility and uselessness of it is irrelevant to the need to communicate.