A wave of research improves reinforcement learning algorithms by pre-training them as if they were human. Read more at QuantaMagazine.org. Music is \u201cQuasi Motion\u201d by Kevin MacLeod.