Vector Faculty Affiliate Talk - Nisarg Shah

Nisarg Shah is a Faculty Affiliate at he Vector Institute, an Associate Professor of Computer Science at the University of Toronto, and a Research Lead for Ethics of AI at the Schwartz Reisman Institute for Technology and Society. His research develops theoretical foundations for algorithmic fairness across a range of domains including voting, resource allocation, matching, and machine learning.
On April 30, 2024, Nisarg gave a virtual talk on, "Building Foundations of Fairness in AI via Social Choice."
Abstract:
With ubiquitous adoption of AI for decision-making, there is a growing concern about whether various (groups of) individuals are treated fairly. A nascent literature within ML has formulated mathematical notions such as demographic parity and equalized odds to capture what fair decisions should look like. However, I will argue that these notions suffer issues such as limited applicability and sensitivity to group specifications. I will then highlight an emerging literature which studies novel mathematical criteria for algorithmic decision-making, building on ideas of fairness rooted in social choice theory. I will show how these notions circumvent some of the aforementioned issues and broadly apply to problems such as classification, clustering, reinforcement learning, recommender systems, and peer review.