Finding Common Ground—and Synthesizing It

Title: Finding Common Ground—and Synthesizing It Speaker: Nisarg Shah, Associate Professor in the Department of Computer Science at the University of Toronto Abstract: Online deliberation platforms invite people to propose and vote on ideas in search of common ground across society’s fault lines. This talk explores how computation can identify ideas that bridge disagreements and guide large language models (LLMs) to synthesize novel ones with provable guarantees. I will first present a new framework for measuring an idea’s bridging value from approval and disapproval votes alone. It yields two measures which capture how well an idea bridges a population’s many overlapping divides. One has now been used by Change.org’s Civic Dialogues platform in several U.S. deliberations. But what if none of the proposed ideas are bridging enough? I will then present a mathematical model of how LLMs produce novel ideas from existing ones. The model yields algorithmic insights on how LLMs can be used to build consensus systematically. Under an empirically testable condition, these algorithms can provably find ideas unifying a population. We engineered a system which largely meets this condition in civic deliberations. I will discuss experimental results and a pilot deployment on the same Change.org platform. Based on joint work with Carter Blair, Benjamin Cookson, Jakob de Raaij, Ariel D. Procaccia, Maxon Rubin-Toles, Michelle Si, and Serena Wang. Speaker Bio: Nisarg Shah is an Associate Professor of Computer Science at the University of Toronto. He is also a Research Lead for Ethics of AI at the Schwartz Reisman Institute for Technology and Society, a Faculty Affiliate of the Vector Institute for Artificial Intelligence, and an Advisor to the nonprofit AIGS Canada. His distinctions include the Computers and Thought Award (IJCAI, 2024), the Kalai Prize (Game Theory Society, 2024), Innovators Under 35 (MIT Technology Review Asia Pacific, 2022), AI's 10 to Watch (IEEE Intelligent Systems, 2020), Victor Lesser Distinguished Dissertation Award (IFAAMAS, 2016), and PhD Fellowship (Facebook, 2014-15). His research develops theoretical foundations for algorithmic fairness across a range of domains including voting, resource allocation, matching, and machine learning. He has co-developed the not-for-profit website Spliddit.org, which has helped more than 250,000 people make provably fair decisions in their everyday lives. He earned his PhD in computer science at Carnegie Mellon University and was a postdoctoral fellow at Harvard University. The Computer Science Colloquium Series will be held weekly on Thursdays at 2:30pm in SEC LL2.224. There will be refreshments before the talk at 2:15pm.