Date: Wednesday, Dec 6th, 2023
9:00 am – 10:00 am Pacific Time
12:00 pm – 1:00 pm Eastern Time
Location: Weekly Seminar, Zoom
Title: Examining the validity and fairness of predictive AI used for societally high-stakes decision-making
Abstract:
Predictive algorithms are used for decision-making in societally high-stakes settings from child welfare and criminal justice to healthcare and consumer lending. Concerns around the suitability and equity of these algorithms require urgent attention. Much of the current discourse on responsible use focuses on fairness and ethics, often overlooking first-order questions of validity. In this talk, we explore the important role validity plays in responsible use and consider its implications for fairness.
Drawing on validity theory from the social sciences, we develop a taxonomy of challenges that threaten validity in the algorithmic decision-making context. We delve into a couple common challenges — selection bias and missing data — in two societally consequential domains, consumer credit lending and child welfare screening. We illustrate how failure to properly address these issues can invalidate standard fairness assessments and undermine fairness corrective interventions. To resolve these issues, we present an alternative method for conducting fairness assessments and corrective interventions that addresses common forms of selection bias and missing data using techniques from causal inference. We conclude by considering the broader question of governance of high-stakes decision-making algorithms.
Bio:
Amanda Coston is a Postdoc with Microsoft Research in the Statistics and Machine Learning Team. Amanda earned her PhD in Machine Learning and Public Policy at Carnegie Mellon University, and she holds a BSE in Computer Science from Princeton. She will join the Department of Statistics at UC Berkeley in fall 2024 as an Assistant Professor. Her research investigates the validity, equity, and governance of data-driven algorithms used in societally consequential settings.