The Importance of Fit to Purpose in AI Safety

On Season 4 of the Theory and Practice podcast, hosts Anthony Philippakis and Alex Wiltschko explore the many aspects of what it means to be human in the new era of artificial intelligence: from communication to robotic surgery and decision-making.

In episode 1, Dr. Greg Corrado, Distinguished Scientist and Head of Health AI at Google Health, explains how to responsibly introduce AI into healthcare. AI has proven itself in detecting diabetic eye disease, managing the risk of cardiovascular disease, and even encoding medical knowledge to answer patient queries, among many new and exciting applications.

Greg discusses safety concepts in AI: bias, robustness, transparency, explainability, and groundedness. He also discusses developing and maintaining datasets reflecting real-world patient realities and values. 

Following this conversation, Anthony and Alex discuss Brian Christian’s book “The Most Human Human.”