Modern day AI machines learn and improve themselves based on ever-increasing amounts of data that humans and machines generate with each passing day. This happens largely through supervised learning, which is a major branch of Artificial Intelligence. But such machines invariably make mistakes. In his talk, Dhar answers the question "when do we trust autonomous learning AI systems?" by utilizing a "trust heatmap” in order to illustrate how the answer depends on two key elements: how often machines make mistakes and the costs or consequences of these mistakes. The larger question for business executives and policy makers is the following: what does it mean to be an executive leader in the emerging AI era? What is a useful way of thinking about the opportunities and risks associated with AI systems in business and government? What should be your expectations from AI-based prediction systems in your business? Dhar provides answers to these questions.