Auditing Predictive Models for Intersectional Biases
This addresses fairness issues in AI systems for marginalized subgroups, offering a novel auditing tool, though it is incremental as it builds on existing subgroup fairness methods.
The paper tackles the problem of intersectional biases in predictive models, where group fairness criteria may not guarantee fairness for subgroups at the intersection of multiple protected classes, and proposes Conditional Bias Scan (CBS) as an auditing framework, showing it detects previously unidentified biases in COMPAS with higher detection power compared to similar methods.
Predictive models that satisfy group fairness criteria in aggregate for members of a protected class, but do not guarantee subgroup fairness, could produce biased predictions for individuals at the intersection of two or more protected classes. To address this risk, we propose Conditional Bias Scan (CBS), a flexible auditing framework for detecting intersectional biases in classification models. CBS identifies the subgroup for which there is the most significant bias against the protected class, as compared to the equivalent subgroup in the non-protected class, and can incorporate multiple commonly used fairness definitions for both probabilistic and binarized predictions. We show that this methodology can detect previously unidentified intersectional and contextual biases in the COMPAS pre-trial risk assessment tool and has higher bias detection power compared to similar methods that audit for subgroup fairness.