Large Language Models and Medical Ethics Education: Insights from Reflections of Medical Students on Their Experiences

Abstract: Recent developments in medical education suggest that large language models (LLMs), such as ChatGPT, may enhance students’ engagement with complex ethical dilemmas. However, no study has empirically examined students’ actual insights gained from using LLMs to explore medical ethics dilemmas, particularly during the preclinical stage of their studies. This study addresses that gap by presenting findings from a thematic qualitative content analysis of 50 preclinical medical students’ reflections on a case-based learning (CBL) exercise. In this exercise, students used LLMs (e.g., ChatGPT, Claude, Gemini) to generate fictional dilemmas and conduct a Principlism-based analysis, after which they provided a human-crafted critique of the LLMs’ arguments. Students subsequently reflected on the advantages and limitations of using LLMs for ethical deliberation. The results indicate that LLMs broadened students’ perspectives and sharpened their critical thinking, particularly through iterative prompts. Additionally, many students emphasized the advantage of LLMs in providing a structured and organized baseline for discussing ethical dilemmas. At the same time, students highlighted limitations in LLM-generated analyses, such as an insufficient appreciation for the emotional dimensions embedded in ethical dilemmas. They also stressed the importance of retaining their own critical thinking skills. These findings may inform future efforts to integrate LLMs into preclinical medical ethics education while also outlining recommended boundaries for their use in the context of preclinical medical ethics education.

Keywords: Large Language Models (LLMs), medical ethics education, preclinical medical students