AI programs that capture a human expert's knowledge as if-then rules, popular in the 1970s and 1980s. Harold Cohen's AARON drew on this rule-based approach.
You could see AARON as a proposal for a combination of more old-fashioned expert systems with today's statistical text-to-image models. And personally, I think that AI will return a bit more to expert systems.
Christiane Paul, Christiane Paul on Curating Cohen’s AARON
He notes a shift from the hard-coded symbolic logic of Cohen’s expert systems to deep learning’s subsymbolic computation—where knowledge is no longer binary or human-readable but instead operates through latent spaces.
Peter Bauman, Christiane Paul on Defining AI Art