@TheMNWolf @susan77 AI generators are essentially applied statistical models, so even calling them "artificial intelligence" is a misnomer. They basically take a bunch of input, find commonalities attached to keywords and other variables, then spit out something that looks statistically like their input.
A statistical model doesn't have any idea what it's doing; it can't have an idea at all because it doesn't think or have any senses.
@TheMNWolf @susan77 Those large language models that can generate large blocks of plausible-looking text are a great example of this.
Statistically, most scientific papers or legal documents look the same, so if you ask it to generate one of those it'll make something that has fake citations and legal cases noted in them. The same thing happens with visual generators as well.
@TheMNWolf @susan77 Anyway, "inspiration" and "learning" are two things they're incapable of doing because they can't do things like free associate, they don't have understanding of their input or any sort of context, and they can't have ideas or curiosity.