Virginia Tech® home

Seminar: Responsible AI by & for the People: Supporting bottom-up remediation of AI harm

Alicia DeVrio

PhD Candidate, Carnegie Mellon
School of Computer Science

Tuesday, March 17, 2026
9:30 - 10:30 a.m.
1100 Torgersen Hall

 

Abstract

AI systems deployed in the world can behave in ways that cause harm, such as by amplifying existing biases and social inequities or creating new ones. Much work in responsible AI has attempted to remediate AI harm, yet these largely top-down approaches have had limited ability to find and address the breadth of potential issues in deployed AI systems. In contrast, my work focuses on how to directly support those negatively impacted by AI systems, who frequently take action themselves to handle harmful AI behaviors they encounter. In this talk, I will discuss my research exploring participatory approaches to AI auditing, showing how the public can surface AI harms overlooked by top-down auditing approaches. I will chart my long-term research agenda extending this work to understand how technologists can better support the ways that people try to remediate AI harm on the ground. As a whole, my research highlights the power of bottom-up remediation efforts and the importance of affected people’s self-determination in working toward AI harm remediation.

Biography

Alicia DeVrio studies how AI systems perpetuate structural harm and advances approaches that aim to restore autonomy and control to those most impacted by these technologies. She is currently a PhD candidate at Carnegie Mellon University’s School of Computer Science and works in the areas of human-computer interaction and responsible AI. Previously, she worked in industry at Microsoft Research, Noom, and the Wall Street Journal and Barron’s at Dow Jones. In addition to publications at top-tier computing venues such as ACM CHI, CSCW, and FAccT, her scholarly work has informed AI policy and is taught in several university courses. Alicia is also a practicing fiber artist who recently was artist-in-residence at Praxis Fiber Workshop’s Digital Weaving Lab.