As far as I'm concerned, I prefer to say "ML generated images" rather than "art", because, well, precisely.
Also: I understand the benefit of ML for *extracting* info from sources, not for creating new material. I trust ML to answer questions like "Based on all scores written in the 17th century, what's the probability that *this* pourportedly 17th-century score is genuine?"; but if I want a "new 17th century-like score", I'll ask a human.