This seems like something DeepMind could create, given the incentive. They were able to train AlphaGo to predict human moves in Go at a very high accuracy (obvi

This seems like something DeepMind could create, given the incentive. They were able to train AlphaGo to predict human moves in Go at a very high accuracy (obviously not with AlphaGo Zero, but the inferior human-predictive version is how they determined that AGZ is playing qualitatively differently).