>A creative brain is a million monkeys with typewriters, plus a subjective critic judging their work. Coming up with random shit and then discriminating among the results is what creative people do

ok but that's not what these models are doing? they're mimicking training data, and sure it's really impressive how far that gets - but ask it to do something outside of that raining data or where the # of samples is low and it's "creativity" falls apart

I think these models exhibit a form of creativity, but not creativity in the vein of say - an art style not seen in the training data