“How do we prepare physicians-in-training to use artificial intelligence (AI) safely and thoughtfully, within evolving clinical and legal frameworks?”
That question, posed by Amith Skandhan, MD, is the heart of new AI training sessions at UT Health San Antonio Long School of Medicine.
Dr. Skandhan is an associate professor and internal medicine hospitalist at the Long School of Medicine. He also serves as a consultant to the Texas Medical Association’s Committee on Health Information Technology and Augmented Intelligence.
Last year, Dr. Skandhan and his university colleagues realized physicians-in-training were already using AI but without formal education or clear ethical and legal guidance. For example, Dr. Skandhan says his students were comfortable using platforms like ChatGPT but did not always know how to do so in a HIPAA-compliant manner.
Some residents were also unaware they needed to comply with AI laws passed during last year’s legislative session when integrating the technology into patient care. Texas Senate Bill 1188 and House Bill 149, for instance, require physicians to disclose their use of AI under certain circumstances – mandates Dr. Skandhan said “add additional layers of responsibility” physicians may not know about.
“AI isn’t waiting for medicine to catch up, it’s already at the bedside,” Dr. Skandhan said. “The question is whether we prepare physicians to use it safely.”
In furtherance of that, the Long School of Medicine began training its internal medicine residents how to appropriately integrate AI into health care via 60- to 90-minute education sessions focused on foundational AI concepts, such as how the technology works, its limitations, and its legal considerations in Texas.
The sessions also trained residents on how to use AI in appropriate clinical scenarios through hands-on prompt engineering exercises, which taught them how to design, refine, and optimize text-based instructions to guide generative AI models to produce useful outputs.
For instance, under the guidance of academic physicians, residents learned how to use and tailor prompts for at least two different types of large language models – like ChatGPT and Gemini – to personalize patient materials such as discharge directions and after-care instructions.
“We did this so residents could see that not every AI tool responds the same way,” Dr. Skandhan said.
After five weeks, Dr. Skandhan says he saw an improvement in residents’ understanding of how to use AI in clinical care. The university presented his findings at the Innovations in Medical Education Conference in Miami in late March – where Dr. Skandhan said he saw a “growing interest” among physicians and medical leaders to move AI education from “awareness of AI to safe, practical implementation.”
Dr. Skandhan believes the safe integration of AI into health care begins at the academic level – and hopes to see more residency programs adopt similar education sessions.
“There are so many questions that come up with AI,” he said. “Physicians need to understand how to use this technology while maintaining the skills they learned in medical school.”
For more information about AI, see TMA’s comprehensive AI resource center.
Alisa Pierce
Reporter, Division of Communications and Marketing
(512) 370-1469