AI Enters Police Work: Goodyear PD Uses ChatGPT to Generate Suspect Images
The Goodyear Police Department in Arizona is pioneering a new approach to suspect identification, leveraging the power of artificial intelligence. The department is utilizing ChatGPT to create images of suspects based on witness descriptions, replacing traditional pen-and-paper police sketches.
The Goodyear PD announced the initiative on social media, expressing hope that “these new techniques and AI technology will assist in solving more complex cases in the future, here in Arizona and around the country.” The department is careful to note that each AI-generated image is explicitly labeled as not depicting a real person, and is “based on victim/witness statements.”
However, reports indicate a nuanced approach to the technology’s implementation. According to the Washington Post, Goodyear PD officers maintain the process isn’t simply an “AI fabrication.” Instead, they emphasize that the AI is used to refine sketches already created through standard investigative techniques – interviewing witnesses and producing an initial drawing, which is then fed into ChatGPT. This practice has raised concerns among experts, who warn it could introduce distortions into an already imperfect identification process and potentially face legal challenges in court.
Leading the charge within the department is Officer Mike Bonasera, the officer responsible for creating the sketches and now overseeing their AI-assisted recreation. He believes AI image generators are a vital tool for modern policing, arguing that “we’re now in a day and age where if we post a pencil drawing, most people are not going to acknowledge it.”
Officer Bonasera, who has five years of experience sketching suspects – requiring specialized training in facial feature drawing, interviewing techniques, and certification – was initially struck by the realism of images generated by ChatGPT when he input his previous sketches. After securing approval from department leaders and the Maricopa County Attorney’s Office, he began using the technology in April. He posits that the hyper-realistic renderings are more likely to resonate with the public, particularly younger demographics, stating, “People are so visual, and that’s why this works.” Currently, Goodyear appears to be the only police department in the United States employing AI for this purpose.
Despite the potential benefits, experts caution against uncritical adoption of the technology. AI image generators like ChatGPT draw upon vast databases of existing images, inevitably reflecting the biases present within those datasets. As Dr. Bryan Schwartz, an associate clinical professor of law, explained to the Washington Post, “That was something we saw early on with some of these generators. That they were really good at creating White faces and not as good at creating some of other races.” This inherent bias could lead to inaccurate or discriminatory outcomes in suspect identification.
