While I agree with you for 2D art, I am using AI to enhance/clean up parts for 3D cgs. It's just so much easier to have the AI do the cleanup than to go in to photoshop and clean it up manually. And also, AI can add in details that make the 3d models look hotter.
For example: without AI
View attachment 3042101
with AI help/clean up
View attachment 3042103
so AI does come in handy but probably not as good for 2D than for 3D.
I by chance know a bit about AI model. The model using for cleanups are totally different from model for generation.
The model for clean up are mostly designed and trained for clean up/adjust a set of features in the vector space of the picture. Those model are already well developed for the past 10-15 years. Examples including the model to remove mosaic, re-focus the picture, filling a missing area in the background.
The model to generate picture are model that become popular or more accurately speaking more advanced with the GAN(Generative Adversarial Network) model which are proposed in 2014.
(both above paragraphs are over simplifying the actual cases, so don't blame me too much)
so if you use a generative model to generate an 2D art, and use good cleanup model to enhance the generated art, you can get really good result.
And I know a lot artist mentioned they start to use new generative model to make something that can inspiring them and be the base of their art. which means they also need to spend quite some time to polish the AI generated picture to make it real art, and this is what I don't like what Rannero is doing. he obviously didn't spend much time to polish the AI generated picture to match his previous manual drawing in almost all aspects: body figure, character outline, the coloring, the shadow etc.