Duke University computer science professor Cynthia Rudin explains why black-box machine learning models should not be trusted in critical settings.
