The reason I don't use AI is because it gives me far less reliable and impossible to specify results than just searching through the limited lists of human made art.
Today, for undisclosed reasons, I needed vector art of peanuts. I found imperfect but usable human made art within seconds from a search engine. I then spent around 15 - 25 minutes trying to get something closer to my vision using ChatGPT, and using the imperfect art I'd found as a style guide. I got lots of "huh that's cool what AI can do" but nothing useful. Nothing closer to my vision than what I started with.
By coincidence it's the first time I'vr tried making art with AI in about a year, but back then I bought a Midjourney account and spent a month making loads of art, then installed SD on my laptop and spent another couple of weeks playing around with that. So it's not like I'm lacking experience. What I've found so far is that AI art generators are great for generating articles like this one. And they do make some genuinely cool pictures, it blows my mind that computers can do this now.
It's just when I sit down with a real world task that has specific, concrete requirements... I find them useless.
Other than stock photos, porn is the killer app for that, but most of the AI companies don't want to allow that.