Study to use AI to analyze LAPD officers' language during traffic stops
Researchers will use artificial intelligence to analyze the tone and word choice that LAPD officers use during traffic stops
By Libor Jany
Los Angeles Times
LOS ANGELES — Researchers will use artificial intelligence to analyze the tone and word choice that LAPD officers use during traffic stops, the department announced Tuesday, part of a broader study of whether police language sometimes unnecessarily escalates public encounters.
Findings from the study, conducted by researchers from USC and elsewhere, will be used to help train officers on how best to navigate encounters with the public and to “promote accountability,” said Cmdr. Marla R. Ciuffetelli of the Office of Constitutional Policing & Policy.
Machine learning, she said at a meeting of the Board of Police Commissioners, “is in its infancy, but will undoubtedly become a profound element in officer training in the future.”
Over three years, researchers will review body camera footage from roughly 1,000 traffic stops, then develop criteria on what constitutes an appropriate interaction based on public and office feedback and a review of the department’s policies, according to Benjamin A.T. Graham, an associate professor of international relations at USC and one of the study’s authors.
These criteria will then be fed into a machine learning program, which will “learn” how to review videos on its own and flag instances where officers cross the line, Graham said.
He acknowledged that certain standards are subjective, and that researchers may not always have the same observations of a single interaction.
“Even something as simple as, did the officer introduce themselves?” he said.
In analyzing the findings, researchers will consider such factors as the location of the stop and the driver’s race, as well the officer’s rank, age and experience. They will also go to “great pains” to anonymize officers and subjects, Graham said.
University researchers from Georgetown, UC Riverside and Texas will also be involved in the study.
Commissioner William Briggs, one of the study’s main proponents, said he was hopeful of the potential but concerned about establishing “safety valves” for increasingly powerful AI systems such as ChatGPT.
In June, President Joe Biden said during a visit to the Bay Area that AI technology comes with risks such as fueling disinformation and job losses, which his administration wants to tackle. Earlier in the year, a group of scientists and executives issued a letter warning that AI could one day pose a threat to humanity, on the level of pandemics and nuclear arsenals. And yet, regulators have been slow to keep pace.
Unlike other large police agencies like New York City’s, the LAPD does not have a dedicated unit to audit the countless hours of body camera footage gathered from police encounters every month. The department does review footage, mostly from incidents in which officers use force or after a personnel complaint has been filed.
Officer rudeness is among the most common complaints from the public. Under LAPD policy, aggressive or profane language is banned, and the department says it doesn’t recognize so-called “tactical language,” to get compliance from uncooperative suspects. But in practice, its use is tolerated in some situations and discipline is rarely meted out for offenders.
At the Academy, LAPD police cadets are instructed that what they say to the public — and how they say it — can affect the outcome of an encounter, with de-escalation emphasized at every turn.
The USC study comes on the heels of similar efforts by researchers at Stanford and the University of Michigan.
Michigan researchers played audio recordings from hundreds of routine traffic stops for participants, who were asked to rate officers’ interactions with motorists, whose race wasn’t revealed. The report found that officers communicated in a more respectful and friendlier manner with white men, while taking a less positive tone with Black men.
Researchers found that the beginning of the conversation — within the officer’s first 40 words, in some cases — was a strong predictor of how the rest of the encounter would go and whether it led to a search or an arrest.
The LAPD’s Inspector General’s office is conducting its own study on officer language use.
During their presentation Tuesday, department officials didn’t say whether the study’s findings would only be used for training or if any misconduct uncovered could lead to discipline — something the police union is likely to oppose.
While saying that it’s “hard to speculate what’s going to happen with the research until it’s done,” Ciuffetelli said the study would likely unfold in four phases. The findings will eventually be incorporated into the department’s “training models,” she said.
Graham said it would likely take six months for researchers to gather the data, and that the preliminary results likely wouldn’t be made available until a year into the study.
Commission Vice President Rasha Shields asked whether a “corollary analysis” would be conducted into the actions of the person being stopped, and how those might also affect the outcome of the stop.
Graham said it would, adding that “most of those questions are about the officers’ speech.”
©2023 Los Angeles Times.
Visit at latimes.com.
Distributed by Tribune Content Agency, LLC.