Nathan Young comments on AI risk/​reward: A simple model