AI Could Make Police Lineups Fairer, Texas Researchers Say

Eyewitness misidentification sends innocent people to prison, but researchers in Texas believe artificial intelligence might help solve this problem.

Dr. Curt Carlson at East Texas A&M University is leading a groundbreaking study on how AI-generated faces could transform police lineups. With $100,000 in funding from The Texas A&M University System, his team is combining psychology, face perception research, and eye-tracking technology to find out if AI can make witness identifications more reliable.

The challenge sounds simple but carries enormous stakes. When police have a suspect, they need to create a lineup that includes innocent people called fillers. If the suspect stands out too much, witnesses might pick them even if they're not sure. If the fillers look too similar, guilty suspects might escape justice.

Traditionally, police search through databases of mug shots to find fillers matching a witness's description. AI could generate those faces in seconds. Just provide a description or photo, and the technology creates several realistic faces with similar features.

But convenience doesn't guarantee fairness. Carlson worries that mixing one real photo with five AI-generated faces could create subtle differences that make the real person stand out. "Someone's got to do the research to see, wait a second, are you putting the suspect in a problematic situation there because he's in there with five AI-generated fillers?" he said.

The research goes deeper than just asking which face people choose. Working with Dr. Dawn Weatherford at Texas A&M University-San Antonio, the team uses special eye-tracking glasses to record exactly where witnesses look and for how long. The technology reveals whether people examine real and AI faces differently, which features attract attention, and how often their gaze moves between images.

This matters because decades of research show people often remember changeable features like hair and clothing rather than stable characteristics like eye shape. Understanding these patterns could help police build lineups that truly test memory instead of accidentally pointing witnesses toward a particular person.

The Ripple Effect

The research reaches beyond the laboratory into real courtrooms and real lives. Seven doctoral students are working on the project alongside undergraduate research assistants, gaining hands-on experience with questions that directly impact the criminal justice system.

Large-scale online studies involving thousands of participants will complement the detailed eye-tracking work. Together, these approaches will give law enforcement practical recommendations about when AI-generated faces help and when they might hurt.

The team includes Dr. Alyssa Jones from Tarleton State University and Dr. Maria Carlson from East Texas A&M, ensuring expertise in experimental design and data analysis. Their work represents a new frontier where cutting-edge technology meets psychology's oldest questions about memory, perception, and truth.

Science is creating tools that could protect innocent suspects while helping witnesses identify actual criminals.