Physicians at Boston-based Beth Israel Deaconess Medical Center are teaching students to use artificial intelligence chatbots to help diagnose cases, The New York Times reported July 22.
Instructors hope medical students will use chatbots like GPT-4 to do curbside consults — where physicians pull a colleague aside to ask an opinion about a difficult case. The idea is a chatbot could perform the same task by providing suggestions and insights.
But some worry that future physicians may come to rely too heavily on AI. Christopher Smith, MD, director of internal medicine residency program at the medical center, told the Times that medical students and residents are using AI, but "whether they are learning anything is an open question. Part of learning is the struggle. If you outsource learning to GPT, that struggle is gone."
Relying too heavily on AI to make diagnoses could be dangerous, Dr. Smith said.
As a curbside consultant, AI chatbots can provide new suggestions to students, but they have to use it correctly. Many students use chatbots like a search engine, which brings little insight, according to the report. Part of their training includes learning how to ask GPT-4 for a consult using phrasing like, "You are a doctor seeing a 39-year-old woman with knee pain," and listing symptoms. They also suggest asking the bot for its reasoning, as they would with a medical colleague.
"It's not wrong to use these tools," Byron Crowe, MD, an internal medicine physician at the hospital, told the Times. "You just have to use them in the right way."