AI in Policing: Capturing Police Performance (Good and Bad) in Routine, Rare, and Novel Events

Police departments produce vast amounts of body-worn camera (BWC) footage, sometimes petabytes worth of data. While this footage contains valuable information for police departments, researchers, and the public, there has been no efficient way of synthesizing the footage to use in an actionable way. Both police administrators and researchers face the same challenge: too much data, too little time, and insufficient human resources to systematically review the footage. Artificial intelligence (AI) can serve as a tool to overcome this challenge by allowing police and researchers to (a) assess officer performance in routine encounters; (b) identify and examine rare outcomes in police-citizen interactions and (c) facilitate the investigation of novel policing events. This chapter explores AI’s potential not only as a time-saving and efficiency-enhancing technology but also as a means of uncovering critical policing behaviors and outcomes that are otherwise difficult to capture. By doing so, we highlight the implications of AI for improved police practice, oversight, and accountability.