Growing concerns are being raised about medical students and trainees who are using AI tools before fully developing their own clinical judgment. This reliance on AI, such as the chatbot OpenEvidence, could lead to a situation where trainees never acquire essential reasoning skills, rather than just losing them.
OpenEvidence, an AI chatbot for clinicians, is actively used by approximately two-thirds of US doctors for queries on symptoms, drug interactions, and clinical guidelines. Medical trainees have also adopted this tool, often at a crucial stage in their learning.
While AI can provide quick and comprehensive answers, some experts suggest that the struggle to independently formulate diagnoses is a vital part of medical training. This process, shaped by failure and uncertainty, is key to building a physician's clinical reasoning. The unchecked use of AI by trainees risks creating supervisors of reasoning before they become reasoners themselves.
A recent study published in Nature Medicine found that AI tools, including those that draw from the latest medical literature, can be less reliable and sometimes less accurate than general-purpose AI chatbots. This highlights a potential issue of misplaced trust in the AI tools currently used by trainees.
To address these concerns, medical schools and residency programmes are encouraged to implement structural changes. This could involve establishing an expectation for trainees to reason independently first, then consult AI, and to periodically work through cases without AI to assess their unaided reasoning.