Reinforcement Learning and Ethics