Mark Bedaywi

Mark is a PhD student working on AI safety and EconCS, advised by Nika Haghtalab and Stuart Russell.

He’s done work on fairly aggregating preferences in multi-agent systems, accelerating reinforcement learning algorithms, and identifying safety failure modes for AI assistants. His current interests include collaborative learning between agents, inverse reinforcement learning and assistance games, and understanding the game theory of persuasion, agreement, and cooperation. Learn more at his website.