However, the study will anonymize LAPD officers and public subjects.

The Los Angeles Police Department (LAPD) has announced a new initiative to utilize artificial intelligence (AI) in analyzing body camera videos to evaluate officers’ language and tone during traffic stops. The goal of this project is to determine whether police language can sometimes unnecessarily escalate encounters with the public. The study, which will span three years, will involve researchers from the University of Southern California (USC) and other institutions.

The research will involve reviewing approximately 1,000 body camera videos from traffic stops and developing criteria for appropriate interactions based on public feedback, officer feedback, and a review of department policies. These criteria will then be integrated into a machine learning program, which will learn to identify instances where officers might be crossing boundaries in their communication. The project acknowledges the subjective nature of these standards and the potential variations in observers’ interpretations.

Researchers will consider factors such as location, driver’s race, officer’s rank, age, and experience in their analysis. The data will be anonymized to protect privacy. Additionally, other universities, including Georgetown, UC Riverside, and Texas, will participate in the study.

The initiative aims to enhance officer training and accountability by incorporating AI-driven insights into interactions with the public. However, concerns about the power of AI, as voiced by Commissioner William Briggs, highlight the need for careful oversight. The study comes in response to ongoing complaints about officer rudeness and the potential racial bias in police interactions. Earlier research from institutions like Stanford and the University of Michigan has shown how initial officer communication can impact the course of an encounter.

The LAPD’s Inspector General’s office is also independently studying officer language use. It remains to be seen whether the findings will solely contribute to training or could lead to disciplinary actions against officers. The preliminary results are expected to emerge around a year into the study, with full integration into training models envisioned in the future.

Related Story:  The DHS Is Using Facial Recognition To Mass Solve Old Child Abuse Cases

Image of LAPD (2010): Flickr