“AlphaGo was largely a scaling up of the principles demonstrated by TD-Gammon, a reinforcement learning program by Gerald Tesauro that defeated the world's best backgammon players.”