Algorithmic UDAAP - Bridging Prediction and Intervention Problems in Social Systems

Explore the application of two distinct legal frameworks for evaluating algorithmic discrimination in this 32-minute conference talk from the Simons Institute. Compare disparate impact (DI) and unfair, deceptive, or abusive acts or practices (UDAAP) as regulatory tools for addressing algorithmic bias, with a specific focus on fair lending practices. Examine how traditional disparate impact law has served as the foundation for fair lending regulation, while recent regulatory efforts have begun invoking UDAAP—a consumer protection doctrine—as an alternative approach to combat algorithmic discrimination. Learn about the formalization and operationalization of both legal doctrines through simulated lending scenarios to assess their effectiveness in evaluating algorithmic disparities. Discover why UDAAP represents an independent and analytically distinct framework from disparate impact, particularly through its "unfairness" prong that introduces concepts of harm avoidability and proportionality balancing. Understand how UDAAP's "deceptive" and "abusive" standards may capture forms of algorithmic harm that escape traditional disparate impact analysis. Gain insights into the unresolved ambiguities that emerge when translating UDAAP principles into algorithmic contexts and the critical need for additional regulatory guidance to establish workable standards for algorithmic fairness evaluation.