S D Med. 2026 May;79(suppl 5):s44.
ABSTRACT
INTRODUCTION: Generative artificial intelligence (AI) tools are increasingly used by medical students and may influence clinical reasoning. These systems can assist with information retrieval, differential diagnosis generation, and clinical decision support. Studies demonstrate widespread student interest in AI despite limited formal training. However, concerns remain regarding overreliance, inaccurate outputs, and effects on diagnostic reasoning. This study evaluated how second-year medical students use AI during clinical reasoning and their primary concerns.
METHODS: We conducted a survey of second-year medical students at the USD Sanford School of Medicine. Students reported frequency of AI use for clinical reasoning tasks, including clarifying medical knowledge, generating differential diagnoses, prioritizing diagnoses, suggesting diagnostic tests, and recommending management strategies. Students also identified commonly used AI platforms and completed an open-ended question regarding concerns. Data were analyzed descriptively, with thematic review of qualitative responses.
RESULTS: A total of 28 students completed the survey. AI use was widespread, with nearly all respondents reporting use. ChatGPT was the most commonly used platform (~90%), followed by OpenEvidence. Students most frequently used AI to suggest diagnostic tests, propose management strategies, and clarify medical knowledge, indicating early integration into clinical reasoning workflows. Common concerns included overreliance on AI, loss of independent diagnostic reasoning, and incorrect AI-generated information.
CONCLUSIONS: AI is already integrated into clinical reasoning among medical students. Educational efforts should emphasize critical evaluation and calibrated use of AI to support, rather than replace, independent diagnostic reasoning.
PMID:42526014