Can AI-Driven Body-Worn Camera Review Improve Police Officer Professionalism? Findings from Three Randomized Control Trials in Arizona

• In Event: Experimental Methods in Criminology (Sponsored by the Division of Experimental Criminology)

Abstract

As body-worn cameras (BWCs) diffused rapidly in policing, departments have struggled with the sheer volume of footage generated every single day. Studies have demonstrated that only a small fraction of footage (5% or less) is ever viewed by anyone. Artificial Intelligence (AI) has emerged as a potential solution to the BWC footage review problem, as departments can dramatically increase their ability to review footage and optimize its value. Unfortunately, there are few empirical studies of the viability and impact of AI-driven BWC review. The current study seeks to fill this gap through a rigorous evaluation of Truleo, a platform that offers AI-driven analytics of BWC audio using natural language processing. We present findings from 6-month randomized control trials conducted with the Apache Junction and Casa Grande Police Departments, as well as the Arizona Department of Public Safety. We primarily focus on a “professionalism” measure generated by Truleo, which assesses an officer’s use of explanation, threats or use of force, and impolite language in every BWC-recorded interaction. The comparative examination of professionalism among RCT study groups sheds light on the potential for AI-driven BWC analytics to positively impact officer behavior during encounters with community members.

Authors

• Mike White, Arizona State University

• Aili Malm, California State University, Long Beach

• Seth Watts, Arizona State University