"Within the next three years, robotics should be completely solved, AI should solve a long-standing unproven theorem, programming competitions should be won consistently by AIs, and there should be convincing chatbots" (2017)

"We will completely solve the problem of adversarial examples by the end of August"