HCII PhD Thesis Proposal - Hank Lee
When
-
Where
NSH 4305
Description
Title: Supporting Practitioners in Addressing AI Privacy Harms
Abstract:
As AI transforms industries by enabling new products and technological opportunities, we are only beginning to understand how it reshapes privacy. My dissertation advances this agenda by developing frameworks and tools that help practitioners identify and address AI privacy harms. I begin by analyzing documented AI privacy incidents to synthesize a taxonomy of risks, showing how AI can both exacerbate existing privacy risks and create new ones. Second, I interview practitioners to examine how they define and scope AI privacy work, what motivates and constrains this work, and what shapes their ability to carry it out. Third, I develop Privy, a tool designed to foreground practitioners’ privacy work during AI product development by helping them identify relevant and severe privacy risks and propose effective, appropriate mitigation plans. Finally, in my proposed work, I will conduct a contextual inquiry with an industry partner to examine principle–practice gaps in practice and investigate whether and how a privacy assistant tool like Privy shapes product teams’ on-the-ground privacy practices. Overall, my research lays theoretical and empirical foundations for privacy-preserving AI innovation while also contributing concrete tools to support practitioners in realizing it.
Committee:
Sauvik Das (co-chair), Carnegie Mellon University
Jodi Forlizzi (co-chair), Carnegie Mellon University
Lorrie Faith Cranor, Carnegie Mellon University
Yang Wang, University of Illinois at Urbana-Champaign
Tadayoshi Kohno, Georgetown University
Zoom link
Link to thesis: https://drive.google.com/file/d/1xotySIPk1-XjMdOMPE0rUbarOLY3EQCA/view?…