tricky_labyrinth comments on The best way so far to explain AI risk: The Precipice (p. 137-149)