Conjecture: Art AI makes art more like science. Individual contributions matter less, the cumulation matters more. That's a big mismatch for the current funding model. We should fund artists well (of course) and encourage them to develop "orthogonal art" — art not within the range of extrapolation from the existing corpus. I'm not sure whether there is a way to measure this, with respect to existing models.

Critique (one of many): Subsumes the expressive dimension of art to the instrumental.