Organizers: Tsvi Benson-Tilsen and Andrew Critch
Time: Tuesdays, 5pm – 630pm
Location: 732 Evans
This seminar will explore mathematical problems faced by artificial agents with the capacity for precise and accurate models of themselves and other agents, and the challenges they pose for the safe and reliable design of such agents. For example, artificial agents might encounter mathematical proofs about their own behavior and the behavior of others, or create new agents to support or replace themselves. How should an agent behave in such scenarios? And given this, how should they be designed?