Summit on AI, Body-Worn Cameras, and the Future of Policing

Can training police officers on how to best interact with the public actually improve their interactions with community members? This has been a challenging question to answer.

In 2023, the team showed they could use the “linguistic signature” of the first 45 words an officer speaks during a car stop to predict escalation and whether the driver would ultimately be handcuffed, searched, or arrested.

Police body-worn cameras have the potential to improve accountability and transparency in policing. Yet in practice, they result in millions of hours of footage that is never reviewed. We investigate the potential of large pre-trained speech models for facilitating reviews, focusing on ASR and officer speech detection in footage from traffic stops.

The research team conducted five studies on officers’ tone of voice in body-worn camera footage from routine traffic stops. The studies revealed that officers spoke in a more positive tone to White drivers than Black drivers, and that community members who listened to clips of officers speaking in a more negative tone placed less institutional trust in the police department from which the interactions originated and had a more negative view of police generally.

The team conducted several follow-up studies using camera footage, modeling the conversational patterns of traffic stops to uncover insights about the gaps and inconsistencies in which police procedural acts occur, which can be used to inform and aid efforts to ensure the procedural justice of police-community interactions.

In 2017, Stanford researchers used AI and NLP tools to analyze large amounts of footage from body-worn cameras at nearly 1,000 routine traffic stops in Oakland, CA. Analysis revealed that officers consistently spoke to Black drivers with less respect than White drivers, a disparity that has important implications for procedural justice and the building of police-community trust.