Title: AI for Societal Decision-Making

Abstract: Large-scale machine learning models trained on human preferences have led to significant advances in AI capabilities. The next frontier of AI lies in building reliable decision-making systems that society can utilize to support consequential decisions. In this talk, I will first present a framework of AI-enabled societal decision-making and provide a brief overview of my research in several stages of the framework ranging from information elicitation to evaluation. The second part of the talk will focus on the preference aggregation stage, and I will outline my research on designing improved voting rules for aggregating diverse preferences. I will conclude by highlighting some exciting research directions at the intersection of reinforcement learning and preference aggregation with applications focused on societal problems.

Bio: Debmalya Mandal is a postdoctoral researcher at the Max Planck Institute for Software Systems in Saarbrücken, Germany. Previously, he was a postdoctoral fellow at the Data Science Institute of Columbia University. He obtained his Ph.D. in Computer Science from Harvard University where he was part of the EconCS group. He works on problems in AI and multi-agent systems, with an emphasis on theory and modeling, and with applications focused on societal problems. His paper has been selected for Oral presentation at the NeurIPS conference and he also received a Certificate of Distinction in Teaching from Harvard.