Medical Programs Urged to Curb Trainees' AI Use as 2-3 of US Doctors Rely on OpenEvidence
Updated
Updated · The Guardian · Aug 10
Medical Programs Urged to Curb Trainees' AI Use as 2-3 of US Doctors Rely on OpenEvidence
2 articles · Updated · The Guardian · Aug 10
Summary
Medical schools and residency programs are being urged to require trainees to reason through cases before consulting AI, to avoid “never-skilling” before clinical judgment is fully formed.
About two-thirds of US doctors already use OpenEvidence, and trainees can use it to generate polished differential diagnoses that mask gaps training is supposed to expose through struggle and supervision.
A recent Nature Medicine study found literature-linked tools like OpenEvidence can be less reliable than they appear and in some cases less accurate than general-purpose chatbots, sharpening concerns about misplaced trust.
Proposed fixes include written pre-AI assessments, rounds that pause before AI consultation, periodic no-AI cases, and drills that teach trainees to spot both flawed and correct AI outputs.
The broader aim is to keep AI as a tutor rather than a substitute for bedside reasoning, so future doctors can detect when a machine is wrong, incomplete, or right for the wrong reason.