A Low-cost Ethics Shaping Approach for Designing Reinforcement Learning Agents

• Yueh-Hua Wu and Shou-De Lin.
• authors investigated how to enable RL to take ethics into account ethics shaping
• assuming that the majority of observed human behaviours are ethical, the proposed approach learns ethical shaping policies from available human behaviour data in given application domains
• rewards positive ethical decisions, punishes negative ethical decisions, and remains neutral when ethical considerations are not involved