Agreed. Symbolic AI dominated the field for 50 years: thanks partially to an accident of history (the highly influential work of Minsky, Newell, etc) and parti

Agreed. Symbolic AI dominated the field for 50 years: thanks partially to an accident of history (the highly influential work of Minsky, Newell, etc) and partially due to lack of the data needed to try anything else. Now we’ve seen essentially 50 years of research on data-driven methods compressed into 5. In retrospect it makes sense that applications to which symbolic reasoning is especially ill-suited (including artistic rendering) would be the lowest-hanging fruit.