Art is the process and the experience, not the object, so AI results in shallow art
My wife and I were talking about what "art" is. I forget the context, but it comes up from time to time since we both have far too many hobbies.
We have a friend who falls hard into the "what I say is art is art" camp, and who will triumphantly point at a trash can and yell "ART!" at us. It used to annoy me. It's very much a devil's advocate gambit, and the devil rarely needs the help. To be fair, though, I couldn't actually articulate what I thought was the differentiator. Until recently.
The problem I have with AI art
Let me preface this: I may speak in absolutes here, but there's nuance on all sides. So before you get your ruffles twisted thinking I'm being Very Declarative About Reality, just chill and remember that few things are ever black and white.
When I first played with Midjourney, I was firmly in the "prompting an AI is a valid way to make art" camp. I was astounded by what it could do, even then. I think I still sort of agree with that, in principle. But the results are hollow and uninteresting.
These models were trained on millions of pieces taken from millions of artists, and what comes out is a sanded-down average of all of it. What's missing is the sheer weight of experience an artist brings to the work.
The point is the experience. The point is the learning. The point is the craft.
Art, for me, is a verb, not a noun. An artist takes all of her life experience, all of her learning, and all of her craft, and focuses it through some amount of intent to build A Thing that holds a measure of all that. The object at the end has the potential for art. But as a static thing sitting there, it isn't art, at least not the way I think of it.
What makes it art is the experience the observer brings to that object. My life, my understanding, my capabilities, whatever baggage I'm carrying: that's what collides with what the artist put in. That collision is the art.
(Intent x experience x skill) of Artist / Experience of Observer = Art
My experience reading Maus is going to be wildly different from that of someone who survived the Holocaust. Both valid but two separate works of art, in a sense.
AI makes hollow art
The prompter acts as the director of intent for generated art. That intent matters. Since I count it as part of the equation, I guess I have to admit that prompting can produce something that results in art? The caliber is just different.
So, yes, I guess I consider AI generation sort of valid. An AI-generated image put in front of me can still produce an artful experience. But knowing it was generated diminishes the effect significantly. It isn't zero, but it approaches zero. Strongly anti-AI art folks will probably have a demonstrably worse experience, given how strongly they feel. Even then, I'd argue there's still a non-zero artful experience.
Because art can provoke disgust and still be entirely valid. I abhorred Goya's Saturn Devouring His Son for years, and now I use it constantly in my own work (see here and here). But even when I hated it, I'd have 100% called it art. My experience matters, and whatever shape that experience takes, revulsion included, is still the art of it.
So with AI, the machine brings no experience and no craft, and the only intent in the room is a prompt. What you get is an object with very little artistic potential. People will still bring their own experience to generated work, and some of them will have artful experiences with it. But I'd wager the quality of that experience, subjective as it is, pales next to what other objects can offer.
Art is experiential gestalt
Art is the wobbly soup of all these things together. The object is the bowl. The experience is everything in it. AI still makes a bowl. It just can't hold much soup.
Shallow soup cools quickly.