X_Larsson:
The thing is, that right now, AI has usable and effective ways of processing illumination, skin texture, human skin tones, outlines typical structures if object s common around us, but it does not undesrstand context. Proportions AND details get distorted or misconfigured.
Then the whole creation is spoiled. Many (most?) of the AI pictures that I have seen, and that contain humans, have some sort of eerie feeling of "wrong" about them.
I could write more here, but I will refrain from it.
Well, it’s new and evolving fast and the barrier to entry is super low (none of those are bad things. Democratizing self expression is a great thing), so we end up seeing a LOT of bad art made with it. The better stuff - made with newer, more expensive software, by more sophisticated users - doesn’t have the messed up hands and uncanny valley stuff. You can still ask the question of whether it’s art or not. I don’t have that answer.
Another part of the AI problem is how easy it is to share mistakes. A traditional artist who messes up the proportions on a figure or fucks up the hands might not share that one. Some artists are self conscious about sharing any of their work. The stakes with AI are low.
Photoshop came out 30 years ago and it’s still controversial (sometimes for good reason, but as more of a stand-in for fights about beauty standards. It’s not photoshop’s fault that pictures get airbrushed).