skeptical_lurker comments on “Human-level control through deep reinforcement learning”—computer learns 49 different games