Fort Worth police officers are testing an artificial intelligence training system developed at the University of Texas at Arlington to practice potentially tense encounters and receive real-time feedback on how they communicate and attempt to de-escalate situations.
The technology can simulate interactions with people who may be autistic, otherwise neurodivergent or have disabilities requiring officers to adjust their approach, KERA reported.
UTA has confirmed that Associate Professor Shuchi Deb has begun testing the system with the Fort Worth Police Department, using real-life scenarios and prompts based on principles of de-escalation.
The artificial intelligence can change its behavior depending on how officers handle the interaction.
“We can create lots of different difficulty levels for the officers who are training,” Deb told KERA. “Some of the AI can be very compliant, some of them are not that compliant, some of them can be impossible to de-escalate.”
AI Scores Officers In Real Time
The system does more than simply generate a virtual person for officers to interact with.
The AI reportedly evaluates what an officer says during the exercise and scores individual statements as positive or negative.
Officers can see the assessment through their headsets in real time and receive a more detailed analysis after the exercise concludes.
UTA said the technology can recognize when an officer is listening, being kind or identifying something the simulated person wants to discuss. It can also recognize when an officer fails to establish a connection and adjust the interaction accordingly.
That allows a scenario to change based on an officer’s choices rather than following a predetermined script.
Fort Worth Police Officer Jorge Lopez worked with Deb to develop the training.
“With AI, it takes out the need for a physical human role player, it takes out safety considerations because there’s no physicality to it,” Lopez told KERA. “So, it’s something that we can do a lot more often in training.”
Training For Unpredictable Encounters
The technology is intended in part to prepare officers for situations in which a person’s behavior may stem from autism, neurodivergence or another disability that is not immediately apparent.
During a traffic stop or other interaction, an officer may not know beforehand whether someone requires a different communication approach.
The simulations give officers an opportunity to recognize behavioral cues and practice adjusting their responses without putting an officer, trainer or member of the public at physical risk.
UTA said the system gives officers “perspective and practice” for high-pressure encounters while allowing the AI to react dynamically to the officer’s conduct.
The approach differs from police uses of artificial intelligence focused on surveillance, facial recognition or identifying suspects. Here, AI functions primarily as an interactive training tool.
DOJ Funding Supports Fort Worth Project
The Fort Worth police work is sponsored by the department and funded through the U.S. Department of Justice under Grant No. 15PBJA-23-GG-06172-NTCP, according to UTA.
UTA identified Fort Worth Police Sgt. Nathan Owens as project manager and Lopez as an instructor and designer of training content.
The police program is related to Deb’s broader research into artificial intelligence as a teaching tool, although a separate $750,000 National Science Foundation grant supports her team’s development of an AI conversational tutor for students.
Deb and UTA Associate Professor Mohammad Islam are working with other researchers to develop that educational technology.
Rather than simply supplying students with answers, the system is designed to ask questions, guide users through problems and analyze their performance.
The police application takes a similar interactive approach but applies it to decisions officers must make during potentially high-pressure encounters.
Fort Worth Expands Technology And Training
The project comes as Fort Worth Police Chief Eddie Garcia has emphasized both technology and officer training during his tenure.
The Dallas Express previously reported that Garcia identified technology expansion as one of his priorities during his first months leading the department, alongside staffing and accountability.
DX has also closely followed separate training initiatives within the department.
In July, FWPD announced First Amendment refresher training after acknowledging that an officer made inaccurate statements during a June encounter with Christian street preachers outside Trinity Pride Fest.
Garcia later said officers were “wrong” in how they described the limits of lawful speech during the encounter.
That First Amendment training is separate from UTA’s AI de-escalation project, which focuses on communication and decision-making during simulated interactions.
AI Becomes A Virtual Training Partner
The Fort Worth project demonstrates another way law enforcement agencies may employ artificial intelligence as the technology becomes more widespread.
Instead of using AI primarily to analyze evidence or information after an incident, the UTA system creates encounters before they happen.
Officers can practice dealing with cooperative subjects, people who are difficult to communicate with and scenarios in which successful de-escalation may not be possible.
They can then see how the system evaluated their responses and repeat the exercise without requiring a human role player.
For Fort Worth police, that means artificial intelligence is not just becoming another technology used in policing. It is becoming a tool for preparing officers for unpredictable human interactions they may encounter on the street.
Provided by Dallas Express
