Study Finds AI Scribes Barely Shift 104 Medical Students' Note Quality
Updated
Updated · Yale School of Medicine · May 26
Study Finds AI Scribes Barely Shift 104 Medical Students' Note Quality
7 articles · Updated · Yale School of Medicine · May 26
104 first-year medical students produced generally high-quality clinical notes, and adding AI-generated drafts led to little overall change in documentation quality or clinical reasoning.
47 students who revised their notes with AI saw one notable drawback: 'Chief Complaint' scores slipped slightly because AI drafts often omitted symptom duration.
Lower-performing students improved with the hybrid approach, while survey responses described AI notes as concise, useful starting points that still missed important details.
A minority of students warned that relying on AI could weaken note-writing skills, prompting researchers to recommend safeguards as schools integrate scribes into training.
The authors said future research should test long-term effects in real clinical workflows, where AI notes may be used as initial drafts and could increase automation bias.
Is AI creating more efficient doctors or just better data entry clerks with medical degrees?
When an AI's medical error harms a patient, who faces the legal consequences?