Well, there's nothing new in this per se, text theory suggested a quasi autonomous "weaving" of texts and imaginary content some 60+ years ago. (Well, it's all about semantic fields and the implications were well known. – BTW, Neal Stephenson wrote an hilarious parody on this as an academic mainstream phenomenon in his Cryptonomicon, look up "Text at War".) But there was still some agency and some effort involved, maybe even reputation, some capital, which served as regulating factors. Now it feels more like marketing blurb without a product and no careers to be pursued behind this.

Also, crucially, those representing the demand for AI art are not those who are meant to consume it. It's still too soon for any systemic feedback, other than economic factors and incentives.