A Framework to Leverage Large Language Models to Address Pressing Issues in US Academic Medicine

Academic medicine grapples with multifaceted challenges that demand timely, data-driven planning. Large language models (LLMs) can serve as rapid “thinking boards,” enabling educators to explore solutions more efficiently while retaining critical human judgment. This study demonstrates a transferable framework by using orthopaedic surgery’s long-standing representation gaps, where, for example, Women, Black or African American trainees, and Hispanic/Latine trainees represent only a small portion of orthopaedic residents, as a use-case. OpenAI’s ChatGPT (GPT-4, version 0613) was queried for best practices in balanced representation, belonging, equitable access, mentorship, and collaboration. Eleven content experts iteratively refined prompts on a five-point Likert scale and supplied qualitative commentary to identify omissions and biases. Mean agreement across five thematic tables was 4.5/5, and reviewers endorsed an AI-derived task-force checklist and metric set for monitoring progress. Panel feedback highlighted areas requiring local context, such as “reverse mentoring” and resource allocation, highlighting that LLM outputs must be filtered through stakeholder expertise. The findings illustrate how coupling LLM synthesis with structured human oversight can shorten the time to actionable plans while preserving rigor and inclusivity. Educators in any specialty can adapt this framework to accelerate curricular reform, crisis response, or other complex initiatives without relinquishing control to the algorithm.