How Do We Build Responsible AI? Towards Fair and Participatory AI Designs

Speaker: Vijay Keswani (Duke University)
https://vijaykeswani.github.io/

Details: Apr 8th (Tue) | 4 PM | Online [Teams Link]

Abstract: As the capabilities of AI have expanded, reports of societal and personal harms related to its use have surged. Examples range from systemic biases in AI tools used in healthcare and social media to stereotype propagation in AI-based search and summarization models. In this talk, I will discuss some of my work on methods to audit and mitigate these biases in AI systems. Building unbiased AI tools presents technical challenges (e.g., constrained sampling and optimization) and practical challenges (e.g., limited group information in real-world settings). Through the use case of search and summarization, I will highlight the challenges of addressing representational biases in search results and our socio-technical approaches to mitigate them. In addition to fairness, this talk will emphasize building participatory mechanisms. I will demonstrate how user and stakeholder participation can serve as an effective mechanism to discover and address social harms due to AI and demonstrate its effectiveness in auditing and mitigating biases in image search results.

Bio: Vijay Keswani is a Postdoctoral Associate at Duke University. His research interests center around community-focused AI development and the ethics of data and technology. His work leverages tools from various disciplines to build robust AI models, combining computational and statistical learning mechanisms with methods from law, philosophy, psychology, and economics. He received his PhD from Yale University in 2023. While at Yale, he was also a Resident Fellow at the Information Society Project during 2022-2023 and a 2022 Policy Fellow at the Yale Institute for Social and Policy Studies