Privacy-Preserving Race/Ethnicity Estimation for Algorithmic Bias Measurement in the U.S
Saikrishna Badrinarayanan, Osonde Osoba, Miao Cheng, Ryan Rogers, Sakshi Jain, Rahul Tandra, Natesh S. Pillai
TL;DR
This paper tackles the challenge of measuring AI fairness by race/ethnicity for U.S. LinkedIn members under strict privacy constraints. It introduces the Privacy-Preserving Probabilistic Race/Ethnicity Estimation (PPRE), which combines BISG-derived probabilistic demographics, sparse Self-ID data, and privacy-enhancing technologies (secure two-party computation and differential privacy) to enable aggregate fairness analyses without assigning individual race/ethnicity labels. The method supports multiple disparity estimators and demonstrates a secure computation workflow that preserves privacy while yielding useful fairness metrics, including sample-based FPR disparities and equity checks. The work provides practical guidance on implementation, benchmarking, and governance, and discusses limitations and avenues for extending privacy-preserving fairness testing in large-scale systems. Overall, PPRE offers a viable pathway to monitor and improve AI fairness within an operational platform while maintaining member privacy and control over sensitive attributes.
Abstract
AI fairness measurements, including tests for equal treatment, often take the form of disaggregated evaluations of AI systems. Such measurements are an important part of Responsible AI operations. These measurements compare system performance across demographic groups or sub-populations and typically require member-level demographic signals such as gender, race, ethnicity, and location. However, sensitive member-level demographic attributes like race and ethnicity can be challenging to obtain and use due to platform choices, legal constraints, and cultural norms. In this paper, we focus on the task of enabling AI fairness measurements on race/ethnicity for \emph{U.S. LinkedIn members} in a privacy-preserving manner. We present the Privacy-Preserving Probabilistic Race/Ethnicity Estimation (PPRE) method for performing this task. PPRE combines the Bayesian Improved Surname Geocoding (BISG) model, a sparse LinkedIn survey sample of self-reported demographics, and privacy-enhancing technologies like secure two-party computation and differential privacy to enable meaningful fairness measurements while preserving member privacy. We provide details of the PPRE method and its privacy guarantees. We then illustrate sample measurement operations. We conclude with a review of open research and engineering challenges for expanding our privacy-preserving fairness measurement capabilities.