I’ve been reflecting a lot on this Erick Hoel’s piece. Erick Hoel’s main argument is that art is an act of communication, a special act of consciousness in the production of something beautiful. That art conveys a certain emotional connection, that art, in one way or another conveys the intentionality of the artist, where the intrinsic purpose transcends beyond the material. For Erick Hoel, art depends on a human. Without the human, there’s no consciousness, nor emotional connection. It’s therefore merely beautiful, or pretty, but NOT art.
Recently, a close friend brought to my attention that this argument is weak. That it’s possible for one to get an emotional reaction from AI-generated art–even if these emotions are of disdain. That it’s entirely possible for a computer to absolutely, out of its training data, conjure such a purpose, in which a human will find meaning. AI artwork is not infallible to the human phenomena of projecting meaning to things. It’s what we do. I think I agree, I think throughout history humans have found such an intrinsic purpose in things they had no business producing. Stars (cue the Northern Lights), cloud patterns, and various nebulas. I’m sure biologists call the arrangement of some flowers art, and so do chemists when they think of the careful arrangement of various crystals and molecules. I think when you look at a Contra Luz Opal or a Marine Bioluminescence you agree that this is art even though we had nothing to do with such creation. I doubt there was intentionality behind it. They’re merely facets in which we humans find some sort of communication, even though there’s no evidence that someone somewhere is trying to.
Part of what I find interesting about art is its otherworldliness. It’s a creation only a superhuman is able to churn out. A great artist is willing to go all the way to the impossible, to see what’s beyond, and to bring perspectives/objects/feelings previously unimagined. This doesn’t discount any other reproduction–in painting, or photography of things that exist–but more often than not, what I’ve found impressive, even in reproduction, is when art is able to find something "out of this world" and bring it to light. A lot of natural phenomena fit these criteria, and they weaken Erick Hoel’s argument–that art is incapable of existing beyond its human creator or their intentionality. So, why don’t I consider AI-generated art art?
I think there are two main reasons why. First, I think AI is very good at generating random bullshit, and because of the incredibly cheap computing power, AI is capable of generating A LOT of bullshit. Walter Benjamin would’ve been surprised. It’s basically a monkey on a typewriter typing infinitely. Part of what makes art art is its scarcity. Paintings, even in their similarity of styles and themes, are very rare. One is not able to simply concoct a billion paintings of the same style and theme, then pick a thousand that are good. Even if one would be able to, no it’s simply not possible. The slowness and carefulness of producing a final art form makes art art. The line consideration, the curves, a brush here, a stroke there. The pausation, and the next day’s continuation, and a new direction to take the painting. Any change a human makes, they do so carefully, knowing that they get very few chances to change the direction of what they’re creating and still come out with something impressive.
With nature, it’s maybe a bit more complicated. I know nature is able to produce a lot of random bullshit, and I know this because I’ve been following astronomy from a distance. Random stars and constellations appear all the time. From what I’ve heard, the James Webb Space Telescope is discovering a lot of gems in a shit tonne of ‘bullshit’. I think what’s unique about what nature has produced though, is that they are spread out across lightyears and light centuries of space, effectively making them scarce. I think this scarcity gives art some legitimacy. That when we stumble upon something otherworldly, and rare, then we don’t expect to stumble upon some other similar thing unless we’re really lucky! Surely in nature, we don’t get to erase a parameter or modify a text prompt to get something similar, and/or different from what we came across.
Second, I think with what I consider art, I am able to acknowledge at least a two-way interaction between art and the artist. With nature, we see it striving for balance. For survival. As nature goes through various motions in striving for stability, art could be formed in the process. More than the aesthetics, the rocks, the flowers, and the stars all gave up or aided something before they attained their current "artsy" form. With artists, it’s a form of release, some sort of communication that Hoel was referencing that the artists experience, leaving art as the product. From what I understand, this is the aura Walter Benjamin was talking about. I doubt this is currently possible with the current forms of AI-generated art. AI is surely able to create or generate something, but I don’t think it’s an attempt at communication, given it creates something from its training. Surely it doesn’t give up any of its training knowledge to reproduce art. What does AI give up in the art creation process?
There’s a long way to go before I could consider AI-generated art as art. What I don’t think is long off, however, is when AI-generated art will start being consumed as art, or a substitute for art. Even more worrying is the point when artists have been discarded in favor of AI, and we cannot find any new data to train AI on, so we have to rely on AI output for new data. Unless we get to AI that can generate ART of course, which in my opinion is highly unlikely to be anytime soon. Time will tell, and I desire to be wrong. But for now, I hope for sympathy for the artists, and especially for the strengthening of institutions that support such fundamental humans. I’m not worried about nature, nature will take its course. It always has.
Some proponents of AI-generated art argue that this skepticism is similar to the naysayers of the early internet days or any form of digitization we’ve gone through in the last twenty or so years. I disagree, none of those technologies enabled generation as much as current AI models do. It was merely a transition of the medium. This current AI wave isn’t, it is an attempt at useful creation and generation.