Regulating Algorithms: What and When

Abstract

The regulation of algorithmic decisions, ranging from advanced credit scoring to employment screening, presents unique challenges in aligning firm behavior with regulatory goals, as well as renewed opportunities to develop the timing and methods of regulation and scrutiny. We propose a framework for algorithmic regulation that emphasizes the importance of temporal stages in the regulatory pipeline: ex-ante (pre-training), ex-interim (post-training but pre-deployment), and ex-post (post-deployment). Regulators can choose both the pipeline stage targeted by the legal rule ("rule timing") and the stage at which compliance is assessed ("scrutiny timing"). We analyze the tradeoffs between different regulatory regimes and highlight how ex-interim rules offer a unique opportunity in algorithmic settings compared to the rigidity of ex-ante rules or the bluntness of ex-post rules, and explore the considerations that guide whether regulators might scrutinize ex-interim rules before or after deployment. Using our temporal taxonomy of rule and scrutiny timing, we situate emerging and proposed AI regulations and outline an agenda for developing tools to regulate AI effectively.