Title: "Machine Learning for Proactive Rights Enforcement: The Case of Labor and Housing Rights"

Abstract: How can data science be used to improve the process of rights enforcement? In reactive rights enforcement, policymakers wait for individuals to 1) recognize a rights violation has occurred, 2) decide to report the violation, and 3) navigate reporting bureaucracies. Reactive rights enforcement can lead to inequalities in rights enforcement. In contrast, the projects we feature involve proactive rights enforcement. Rather than waiting for individuals to ask for help, government agencies can use large-scale administrative data to try to identify entities at high risk of violating individuals' rights. We discuss two research projects, each conducted in partnership with government agencies, to highlight the role of data science and supervised machine learning in proactive rights enforcement. The first project focuses on the rights of tenants living in rent-stabilized units in NYC, partnering with an NYC agency that sends outreach workers to learn about issues with their landlords. The second project focuses on the rights of H-2A guest workers, partnering with a legal aid agency that does worker outreach. Summarizing these two projects, we highlight the promises and perils of data science in this context, including issues of label bias, tensions between predictive and interpretable models, and deployment of predictions.

Bio of the speaker: Rebecca Johnson is an Assistant Professor in the Program in Quantitative Social Science, affiliated with Sociology. Her research focuses on the ethics and law of how government bureaucracies (e.g., housing authorities; K-12 schools) use a mix of data and discretion to decide who deserves help, and how that prioritization impacts inequality.

** Registrants will receive a zoom link one day before the webinar.