The synergistic effect of all the AI's inputs absolutely results in a unique new 'flair', with extensions, reversals, and mash-ups of styles just as in human-made artistic styles.

And AI "builds exclusively on past experience and work of humans" just like any young new human artist equally does. In many cases, you can even tell the different models' outputs apart, not by raw quality or glitches, but by hard-to-describe aesthetic tendencies.

I share your concern on the effect on human artists – both the market for their work, and even their morale, when learning, knowing that decades of practice will still be outproduced by seconds of computation.

But I don't think the genie will be put back in the bottle, by either expansive interpretation of existing copyright law, or even new laws.