Purdue University computer scientists plan to use AI to uncover patterns in criminal behavior and predict recidivism. A proposed project, led by professor Marcus Rogers and assistant professor Umit Karabiyik, would track parolees recently released from prison using smartphones and biometric wearables. The resulting data would be used to train an AI algorithm to find patterns linked to regression of criminal behavior.
- Rogers said the goal is to find opportunities to intervene earlier and assist criminals in their reintegration back into society.
- The plan is to recruit 250 volunteer parolees from Indiana. Half will wear bracelets that monitor health data, like heart rate and stress levels, and agree to share data like a location from their smartphones. The other half will be the control group and not be monitored.
- An AI system would analyze their data in chunks, searching for patterns related to criminal behavior. It would not evaluate the data in real-time.
- The project has received criticism. Liz O’Sullivan of the Surveillance Technology Oversight Project told VentureBeat that there’s scant evidence that AI could "infer emotional state from biometrics."