Dr. Ayşe Başar

What does it take to train a machine to think? If you are Ayşe Başar, it isn’t just a relentless drive, years of research into AI and machine learning, and more than a decade of experience working in industry. Building predictive models also relies on an ability to step into someone’s shoes. “To translate what an expert does into mathematics and code, we basically role-play their experiences to understand how they think,” she says.

One such instance came while Başar and her team were building an AI system intended to improve the capacity for emergency room teams at St. Michael’s Hospital. The team’s goal was to assess whether a patient should be discharged or stay in hospital. Once a preliminary model was in place, Başar’s team spent a morning in the ER being trained alongside a group of medical students, ordering tests, looking at patient histories and talking with a patient recovering from surgery. “Being immersed in their world was fascinating and incredibly helpful in modifying our models to think more effectively,” she says.