@hakui It sounds like he's confusing "surprising emergent results" with autonomy. You can work out a fractal from a simple algorithm on pencil and paper and get results that you could not have predicted, but one one thinks that the 1D Life algorithm ( https://en.wikipedia.org/wiki/Elementary_cellular_automaton ) is autonomous. A computer beats Kasparov at chess in 1997, and the program is written by people that could not beat Kasparov themselves. Speculation before that (e.g., "Goedel, Escher, Bach", an actually good book on this topic) was that it would require actual artificial intelligence to beat a GM at chess, but this turned out to be completely wrong: computers are just really good at some tasks.

It seems like most of the stuff he says about computers is stuff that has a contrary example from the 1970s, but "unpredictable output means autonomy" is really egregious.
rule137.gif