Ken,

> Yes, Cohen and Colton created interesting software that creates
> interesting visual art. But I'm not sure that there is much AI in
> either. Aaron was good at simulating a paint brush when given a
> destination - is that AI?

Cohen's work on AARON was inspired by his visit to the Knowledge
Systems Laboratory at Stanford in the 1970s, and he described the
system as using a collection of heuristics to guide composition.

There's a nice summary of the AARON project in Chapter 3 of the 2012
edited book Computers and Creativity:

https://link.springer.com/content/pdf/10.1007/978-3-642-31727-9.pdf

The chapter by Simon Colton reviews aspects of The Painting Fool, which
started as a constraint-based system. AARON was a direct inspiration.

So the short answer is, yes, classic AI representations and processes
were central to both Cohen's and Colton's early work on creativity in
the visual arts.

> To me there is no doubt today's text to image systems are examples
> of AI.

I don't know the details of DALL-E and its kin, but I have the impression
they're closer in spirit to David Cope's work on musical composition:

https://en.wikipedia.org/wiki/David_Cope
https://bernardmarr.com/emi-when-ais-become-creative-and-compose-music/

His EMI system deconstructed human compositions, identified patterns,
and recombined their elements into new musical pieces.

This early work did not rely on neural nets, but it's always good to look
for common themes that cut across computational paradigms.

Best, -Pat