SGIM24 Highlights: Debate on Medical Misinformation in the Era of Large Language Models

This is a recorded session from the SGIM 2024 Annual Meeting in Boston.

Large language models (LLMs) are machine learning models trained on vast amounts of text-based data to comprehend and create human-like text based on learning pattern recognition.  LLMs are anticipated to reshape medicine, from areas ranging from clinical diagnosis to population health.  But LLMs also present challenges, notably the risk of propagating misinformation.  False or misleading information has led to preventable deaths, affected racial/ethnic minority groups disproportionately, and eroded trust in once-respected health sources.  As LLMs become integral to healthcare, addressing its misinformation potential is essential for maintaining trust and ensuring community well-being. Given the trust patients place in primary care physicians, general internists can play a pivotal role in countering this challenge.

1. Discuss the implications of LLMs on the propagation and management of scientific misinformation
2. Describe the challenges and pitfalls when utilizing LLM tools to identify and address misinformation
3. Identify strategies to vet LLM-driven health information outputs, especially in scenarios where evidence is ambiguous or evolving
4. Describe the evolving roles of general internists, medical journals, and LLMs in navigating, defining, and mitigating the risks of misinformation

To receive CME and MOC credit, you must (1) watch the entire presentation, (2) complete the brief MOC Assessment survey, and then (3) click the Certificate item to claim your certificate (print/save for your records).