Having it learn on human games was just a way of speeding up the initialization process before running reinforcement learning, it didn't limit the state tree th

Having it learn on human games was just a way of speeding up the initialization process before running reinforcement learning, it didn't limit the state tree that was being searched later on.