Impact of Democratizing Artificial Intelligence: Using ChatGPT in Medical Education and Training

Abstract

To the Editor: We read the article by Boscardin and colleagues with great interest.1 Although the authors contend that generative artificial intelligence (AI) tools "in their current form cannot be depended on as reliable learning tools,"1 they anticipate opportunities for trainees to incorporate AI in generating differential diagnoses for real patients. We would like to highlight that the recent iterations of generative AI tools, such as ChatGPT and Gemini, have shown their ability to analyze patient cases and provide invaluable insights, like predicting possible disease causes. Such insights can potentially assist students and interns in making informed decisions during their education and training. To assess generative AI's efficacy as a copilot in clinical education, we compared ChatGPT models in preparing teaching contents and systematically benchmarked ChatGPT-4 in symptom checking using hypothetical clinical cases derived from real patient scenarios across a wide range of diseases.2,3 Our findings reveal that ChatGPT has an accuracy rate of approximately 80%. This suggests that the current tools are primed for students to employ as a supplementary aid in clinical case learning. Regarding the integration of generative AI into research, Boscardin and colleagues discussed generative AI tools as sources of information. We would like to expand on this by noting that generative AI holds the potential to democratize AI research for all students and residents. In the past, conducting AI research in health care was confined to a select few with access to AI models and structured data. Today, medical students or interns can leverage publicly available generative AI models to analyze unstructured data within their secure and patient privacy-protected environments. For instance, generative AI tools empower every student and clinician to undertake comparative effectiveness research, evaluating the impact of AI on their specific health care tasks using an embedded research approach of learning health systems. This revolutionary shift is attributed to generative AI's unparalleled prowess in natural language interactions with humans and data.4 Given generative AI's potential as both a learning tool and a research instrument for medical students and professionals, this democratization of AI could profoundly influence medical education and practice. Anjun Chen, PhD Visiting professor, School of Public Health, Guilin Medical University, Guilin, Guangxi, China; email: [email protected]; ORCID: https://orcid.org/0000-0003-4209-8301 Wenjun Chen Medical graduate student, Guilin Medical University, Guilin, Guangxi, China Yanfang Liu, MD Attending physician, Department of Neurology, Guilin Medical University Affiliated Hospital, Guilin, Guangxi, China

Ähnliche Arbeiten