Customer insight, the collective intelligence of teams, lean operations, and value-based imaging…all important topics in radiology today. Still AI and machine learning steal the spotlight every time. Such was the case at the Radiology Leadership Institute (RLI^®) Summit in September, which brought together RLI and one of the nation’s top business schools, Babson College. Interested radiologists and leaders in the academic, private, and government sectors met for four lively days of discussions, many of them on AI and machine learning.

David Louis, MD, chief pathologist at Massachusetts General Hospital in Boston, got us thinking when he explained how pathologists have embraced the strengths of AI in their workflow. Later, we broke into groups to brainstorm the “opportunities” and “threats” that AI and machine learning bring to radiology.

AI Good! — We felt that AI would improve efficiency and precision. And we agreed that AI applications would be useful in measuring and detecting lung nodules, lymph nodes, and multiple sclerosis plaques, as well as detecting findings that aren’t easily seen on conventional scans.

AI Bad! — We wondered how many of us might lose our jobs. Would we have difficulty remaining relevant? With AI taking over some of our normal workload, would we forget how to interpret a normal radiograph in seconds? What happens when a machine learning platform makes an error? Is the radiologist legally responsible? And what would our retraining look like as we become highly trained information specialists? Further, could AI generate hundreds of meaningless incidental findings that radiologists would typically ignore?

AI, It’s Here! — The keynote address was given by Anne LeGrand, general manager of Imaging at IBM Watson Health, who told us that IBM Heath’s goal was to have a machine learning workstation at the desk of every radiologist in the world, making consistent, expert, radiologic interpretation available to all. But not all bad news — LeGrand also expressed that there will be a continued need for radiologists to supervise machine learning platforms, and radiologists would remain relevant.

AI, We’ll Take It!! — A radiologist in the audience summed things up nicely with, “Radiologists that use AI will displace radiologists that don’t.” It seemed a fitting end to the discussion!