CS 239-01 Ethics for AI, Robotics, and Human Robot Interaction

Technical challenges of endowing autonomous artificial agents with normative principles that will allow them to operate successfully in human societies. Algorithmic approaches in artificial agents (rule-based, utility-based, behavior-based, etc.) and their links to philosophical foundations of the main ethical theories (virtue ethics, deontology, utilitarianism). Conceptual and mathematical analysis of assumptions underlying each algorithmic approach as well as implementation in agent-based simulations. Functional and performance tradeoffs of the algorithms and their implications for autonomous robots and AI systems. Mathematical and computational challenges of the different proposals for "implicit" and "explicit" ethical agents, including inverse reinforcement learning, verification-based approaches, and model checking.