The ubiquity of augmented intelligence (AI)—commonly called “artificial intelligence”—has effectively ended the debate about whether physicians in graduate medical education (GME) should use the tool. What remains now is for programs to determine how best to teach health AI literacy, a skill that can improve the care physicians give and bolster their careers when they enter the workforce.
“The conversation a couple years ago was focused on all these concerns around AI. We're not sure that we want to promote the use of AI, therefore, we're not teaching about AI. We've moved past that, fortunately, and now we're in a phase where we recognize that it's indeed those concerns that create the need to teach people,” said Kimberly D. Lomis, MD, the vice president of medical education innovations for the AMA.
“If you're worried that AI may amplify bias or you're worried that residents may be deskilled because of using these tools, the only way to start that conversation is to discuss it with transparency and to provide some training and some fundamentals,” said Dr. Lomis. She helps lead the AMA ChangeMedEd Initiative, which works across the continuum with visionary partners to create and disseminate innovations to better train physicians to meet the needs of patients today and in the future.
The AMA uses the term “augmented intelligence” as a concept that focuses on AI’s assistive role in enhancing human intelligence rather than replacing it.
And AI is everywhere, including in the hands of physician residents and fellows.
Nearly two-thirds of physicians, 66%, surveyed by the AMA (PDF) in 2024 reported using some kind of AI in their practice, a significant increase from the 38% of physicians who said they used it in 2023.
From AI implementation to digital health adoption and EHR usability, the AMA is fighting to make technology work for physicians. That includes recently launching the AMA Center for Digital Health and AI to give physicians a powerful voice in shaping how AI and other digital tools are harnessed to improve the patient and clinician experience. The AMA also recently adopted policy at its Interim Meeting of the House of Delegates to expand training in AI across the medical education continuum.
Major questions on health AI’s use
As residents and fellows join other practicing physicians in using AI as scribes, in research and in a multitude of other applications, residency programs are looking for ways to incorporate the tool into the GME curriculum. The 2024 Work Trend Index Annual Report (PDF) from Microsoft and LinkedIn found that 75% of global knowledge workers—a group that includes physicians—use AI at work, with 78% of those who use AI at work saying they used their own AI. A lack of organizational transparency about tool use raises risks of ineffective use or even the jeopardization of protected health information.
“We’re pretty early on in that journey, as I presume most other health systems are,” said Bruce Levy, MD, director of the clinical informatics fellowship program for the Danville, Pennsylvania-based Geisinger College of Health Sciences. “We're still trying to figure out the professional use of AI, both as a learner and as a clinician, and we're having some great discussions about it, as I'm sure are going on all over the country.”
Geisinger is part of the AMA Health System Member Program, which provides enterprise solutions to equip leadership, physicians and care teams with resources to help drive the future of medicine.
Dr. Levy said it’s critical to understand how AI might affect residents and fellows and shape their skills as physicians.
“How might it impair our ability to ensure that our trainees are appropriately trained? If this tool writes your notes for you, should we be sure that they know how to write a note before they’re allowed to use it? When is the appropriate time to introduce it in training? How much should they use it?” he said. “These are all really important questions that we're going to have to take time to better understand.”
Dr. Lomis said educators also must evaluate each task that may be eliminated or modified and determine whether there is a presumed educational benefit to this task, and, if so, is there an alternate path to accomplishing that educational goal?
Dr. Levy said, though, that he uses AI himself, and believes its correct application must make its way into the GME curriculum.
“I’m using it every day in a variety of ways, and it's great for some things. I don't find it particularly as useful for other things,” he said, noting that ambient listening is one example of a task that generative AI does well and that most physicians find useful. He said it can also be helpful in searching unwieldy documents for specific items and integrating them into presentations, for example. But he said it is not as useful for complex tasks and in longer chats—both as an individual prompt and in a series of prompts.
Dr. Levy added that his experience as a programmer has taught him that the code generative AI writes is “more brute force and less elegant” than that created by humans.
“Certain things you can do and certain things you shouldn’t do” with AI, he said. “Your car can go 100 miles an hour. Should you drive it 100 miles an hour? Probably not.”
Health AI’s a new fact of modern medicine
Residents and fellows, like practicing physicians, both want to and need to use AI in delivering care. Their educators can help them figure out the best way to do that.
Residents and fellows “know how to use specific applications, but they haven't had the fundamentals, in many cases,” Dr. Lomis said, comparing it to someone getting into a car for the first time and driving around a parking lot. “You haven't fully learned how to drive. You've learned how to operate that machine, but you still need the rules of the road and the risks, such as the dangers of texting and driving, or drinking and driving.”
She said it’s vital to teach new physicians how AI is developed, how it is trained, the fallibility of the data sets upon which it is trained, and how the different models function. Programs should collaborate with trainees to design educational interventions in digital literacy, to support and residents and fellows who want to use AI responsibly.
“When I talk to residents about how they're actually using it, it's very encouraging. Very few of them are using it to get a quick answer,” said Dr. Lomis, who recently spoke about the topic at the International Conference on Residency Education, held in Ottawa, Canada. “Most of them are using it to offload administrative-type tasks, similar to how practicing physicians are doing it, or as a sounding board to test ideas and deepen their understanding.”
When it comes to alleviating burnout—an essential topic for all physicians, but particularly for those in residency and fellowship training—AI can be an asset.
“If we can offload tasks that are not useful in terms of the learning part of their educational process, that can free time for other activities, including rest,” Dr. Lomis noted, adding that most residents and fellows are using AI in a “very thoughtful way, not taking away the learning part, but trying to offload some of the burden.”
Dayna Isaacs, MD, MPH, a second-year hematology-oncology fellow in California, said she often uses The New England Journal of Medicine’s AI tool to help find key historical clinical trials for different cancer types.
“Obviously, your experiences with patients and your expertise there supersedes everything,” said Dr. Isaacs, who also is a delegate for the AMA Resident and Fellow Section. “However, we do need to be extremely aware of AI's limitations because sometimes it will make up fake sources and fake publications or just have completely inaccurate data.”
An AMA STEPS Forward® webinar, “AI Tools for Documentation: The Newest Member of the Care Team,” offers additional insight into how ambient AI tools can help make technology work for physicians, care teams and patients.
Also, learn more with the AMA about the emerging landscape of health care AI. And find out how to apply AI to transform health care with the “AMA ChangeMedEd® Artificial Intelligence in Health Care Series.”
