Interesting observations, yet in this essay, the aesthetics of specific models are explored, not art. Art is more than hadicraft. Art is how you explore your inner world and reality with available tools, be it traditional or digital tools. (Sidenote: the models mentioned here are rather Midjourney V 3, which is meanwhile outdated by V 5.2 which can perfect hands, perspectives and emotional portraits).

After all, AI art can be created 1) with use of AI-driven tools (like Photoshop/PremierPro, most of digital tools implement AI meanwhile), 2) in collaboration with AI, 3) unsupervisedly by AI. Also, there are way more artistic approaches than just image2text, AI art is feasible since around 8 years (with DeepDream, GAN, 3DKenBurns, GauGAN etc.), but predicted already by Marcel Duchamp, Umberto Eco, Italo Calvino and more brillant minds of 20th century. It will never replace traditional art, but will come along with it, like photography with painting.