Sequential Decision Making Under Uncertainty - Information Asymmetry in AI Systems

Watch a 53-minute lecture from the Naturalistic Approaches to Artificial Intelligence Workshop at UCLA's Institute for Pure & Applied Mathematics where Max Planck Institute's Rupak Majumdar explores sequential decision-making under uncertainty with information asymmetry. Examine two key problem versions - persuasion and mechanism design - where in persuasion scenarios, better-informed principals influence less-informed agents through information signaling, while in mechanism design, less-informed principals incentivize more-informed agents to reveal information through committed mechanisms. Learn about Markov persuasion and mechanism processes in dynamic models, discovering how these problems can be solved efficiently for myopic agents but become computationally challenging with far-sighted agents.