Tao Lin comments on Human beats SOTA Go AI by learning an adversarial policy