Yeah, what you're saying is true but only to the extent of simulacra AI can produce. But it's not really art if you know what I mean, it's an artefact of training on other people's art.

So what I do with AI art is start with a photo I took. I then extract its Lin Canny lines or other descriptive features algorithmically. From there, I apply various models to re-express the image in an interesting way. IMO it looks more like Digital Art out of photoshop and I would refute any claim that this isn't remotely artistic given I was the one that took the photo that is the base image. And it doesn't end up looking like obvious AI art, but rather more Andy Warholish than anything else IMO.

And sure, the model was trained on other people's art but so are we.

It may not be your cup of tea, but it's a lot more effort than typing in a prompt and calling it a day. But, ya know, Rembrandt and Hans Zimmer were the OG prompt engineers in their respective media IMO.