MiDAS: Automating Unemployment Benefits

In 2015, the state of Michigan considered whether to nominate its Michigan Integrated Data Automated System (MiDAS) for a prestigious state technology award. Launched in 2013 amid severe budget pressures, the $47 million automated fraud detection system was designed to modernize the state's unemployment insurance claims process and reduce improper payments. MiDAS used rules-based AI to autonomously detect and process potential unemployment fraud cases by analyzing datasets and matching claimant information against employer records. The system had exceeded its financial projections—identifying over 25,000 fraud cases in its first year (a 300% increase), generating $96 million in fines, and enabling significant staff reductions.
As Michigan considered the award nomination, several factors required evaluation. The automated process gave claimants 10 days to respond to fraud allegations via an online questionnaire, after which MiDAS would determine fraud and impose penalties of 4x the claimed amount plus interest. The system's rapid implementation had led to a surge in cases and appeals, creating new administrative challenges. Did MiDAS's innovative approach and financial success represent a model for government modernization worthy of immediate recognition? Or was more time needed to evaluate its broader impact?