Washtenaw County Pilots AI That Predicts Residents’ Thoughts Up to Three Seconds Early

Washtenaw County officials on Monday announced a pilot program for a new artificial intelligence system that, according to project materials, can predict a resident’s thoughts up to three seconds before they occur.

The system, dubbed “PreThink,” is being developed by researchers at the University of Michigan’s Center for Predictive Cognition, part of a broader campus initiative focused on anticipatory systems and behavioral analytics. The technology is being tested across several city platforms, including parking meters, the A2 Ahead app, pilot public kiosks, and select university dining services.

County administrators describe the effort as a step toward what researchers call “anticipatory governance,” where services respond to needs before they are fully expressed.

During a demonstration at City Hall, Dr. Elaine Karpov, Director of the Center for Predictive Cognition, approached a PreThink parking kiosk. Using a biometric login, the screen displayed: Yes, you only need 10 more minutes. “I think that should be enough to finish this presentation,” Karpov noted.

In another test, a user opening the PreThink app was immediately shown a notification reading: Yes. Liberty Street is now clear of construction.

At one university dining location, a student test group checks in at a PreThink kiosk and proceeds directly to the cashier without placing an order. Meals arrive within minutes, determined entirely by the system’s predictions.

“We have recently achieved 83.6% accuracy,” said Dr. Marcus Havel, a senior systems engineer on the project. “That will continue to improve as we refine how the system interprets pre-conscious signals.”

“It’s not reading minds in the traditional sense,” Karpov said. “We’re identifying the cognitive subcranial pre-event layer where intent briefly exists before a thought becomes fully formed.”

Currently in a monitored beta phase, the system relies in part on standard smartphone technology to transmit user data to nearby infrastructure, including parking meters and informational kiosks, where predictions are rendered in real time.

According to project documentation, PreThink also incorporates experimental inputs, including faint electromagnetic neural activity detectable through fingertips when interacting with projected capacitive (PCAP) touchscreens. Unlike older technologies that require pressure, PCAP screens detect the minute electrical charge in a bare human finger, allowing researchers to capture neural signals associated with emerging intent.

Optical scans are also used to confirm neural activity patterns during screen interaction. Researchers say these combined inputs help narrow the window between intention and action to a matter of milliseconds. “The old saying is literally true – the eyes are the windows to the soul,” Karpov told the commission.

In addition to neural interpretation, the system processes thousands of contextual data points, including location, time of day, typing cadence, and past behavior, to anticipate and further refine what a person is about to think or do.

Looking ahead, researchers believe the technology could extend beyond convenience and into areas such as medical diagnosis and patient care, particularly in cases where the ability to communicate is limited.

“The body knows what the body knows,” Havel said. “This gives us a way to listen earlier.”

The pilot will run through the summer, with officials evaluating system performance and public response.

While county officials describe PreThink as a step into the future, residents may already recognize a familiar foundation: today’s search engines and smartphones routinely anticipate queries, complete sentences, and surface suggestions based on patterns of behavior. PreThink, for now, simply pushes that idea a little further, and a little faster, than most people are used to seeing.

And if it all sounds just slightly too strange to be true, you’re right. HAPPY APRIL FOOL’S DAY!

Featured image: Dr. Elaine Karpov, Director for the Center for Predictive Cognition, demonstrates the A2 Ahead app on a PreThink kiosk. The advanced AI technology has entered beta testing at several city and university locations. Photo by Taylor Bennett