Date: Thursday, June 13, 2024
Time: 1:30 pm - 2:30 pm
Location: DC 1304
Zoom Link: Click here to join
Passcode: 243770
Title: To Build AI That Works For Everyone, Adapt Models and Coordinate Data
Abstract: As AI products proliferate, questions about whether currently available systems are adequately fair and reliable for all users (and the public more broadly) remain fiercely contested. In this talk, I will suggest some important ingredients for addressing these concerns: model adaptation to specific contexts and distributions, and the collective coordination of users in how they interact with AI systems. I will start by discussing some of my recent work on adapting pre-trained neural networks to better generalize beyond the training distribution. Then I will describe a direction for future work on how everyday people can influence the AI systems they interact with through data coordination. Because machine learning is always data-dependent, I argue for the importance of grassroots collective action in building AI systems that are aligned with community needs.
Bio: Elliot Creager is an Assistant Professor at the University of Waterloo in the Electrical and Computer Engineering Department. He is also a Faculty Affiliate of the Vector Institute for Artificial Intelligence and the Schwartz Reisman Institute of Technology and Society. His research focuses on improving the fairness and robustness of machine learning methods. He completed his PhD at the University of Toronto in 2023 and was previously an intern and student researcher at Google Brain in Toronto.