We humans can learn from DeepMind's Go wins (and loss)