What do the best trials show about AI scribes?
The strongest evidence comes from a randomized controlled trial of 238 physicians across 14 specialties, published in NEJM AI in 2025 [1]. It tested two AI scribe platforms — DAX Copilot and Nabla — against usual care. Nabla users cut their note-writing time by 9.5% (about 2 minutes per visit), while DAX users saw no significant change. Both platforms improved burnout scores: DAX raised the Mini-Z well-being scale by 2.83 points (on a 10–50 scale) and Nabla by 2.69 points, and both reduced physician task load. However, clinicians reported only 'occasional' inaccuracies, scoring around 2.7–2.8 on a 1–5 scale where 5 means 'always accurate.' This means errors happen often enough to require human review.
A second large study — a quality improvement project across six U.S. health systems — followed 272 clinicians who used an AI scribe for 30 days [2]. Burnout dropped from 51.9% to 38.8% — a 13-percentage-point reduction. Clinicians also reported spending 0.9 fewer hours per day on after-hours documentation and improved ability to give patients undivided attention. This study is not a randomized trial, but its size and multi-site design add weight.
A smaller randomized trial in orthopedic surgery (304 patients) found that AI scribes cut physician time in the exam room and administrative processing time significantly [3]. However, it also found 19 medical errors in the AI-scribe group versus 9 in the conventional group — a finding that underscores the need for careful review. The errors were caught by the surgeon before notes were finalized, but the higher error rate is a real concern.
Do AI scribes actually reduce burnout and improve workflow?
Yes, across multiple studies, AI scribes consistently reduce burnout and improve workflow, though the effect varies by platform and user. The randomized trial [1] found that both DAX and Nabla improved burnout scores, task load, and work exhaustion — even though only Nabla significantly cut note-writing time. This suggests that even modest time savings can improve well-being, or that other factors (like reduced cognitive load) matter more than raw minutes saved.
A retrospective cohort study of 125 AI scribe users at an academic health system found that users spent 8.5% less time in the electronic health record (EHR) per appointment and closed encounters 7.1 hours faster [4]. After-hours 'pajama time' did not change significantly, but the overall EHR time reduction was meaningful. A separate pilot at a large academic medical center with 45 physicians found that daily documentation time dropped by nearly 7 minutes and total EHR time by 20 minutes per day [6].
Qualitative interviews with 22 physicians [5] reinforce these findings: 100% of comments about cognitive demand were positive, 62% about temporal demand were positive, and 91% about work-life integration were positive. Physicians said the tool helped them focus on patients and reduced the mental burden of note-writing. However, they also complained about note length and editing requirements, and noted that accuracy with non-English-speaking patients was poor.
What are the caveats and limitations of the current evidence?
The evidence is promising but not yet definitive. The largest randomized trial [1] was a single-site study (though multi-specialty) and lasted only two months. The authors themselves say the burnout findings need confirmation in larger, multi-center trials. The study also had low usage rates — DAX was used in only 33.5% of visits and Nabla in 29.5% — meaning many clinicians didn't consistently use the tool, which could dilute results.
Accuracy remains a concern. In the orthopedic trial [3], AI scribes produced more than twice as many medical errors as traditional dictation. While the errors were caught before finalization, they required extra physician time to correct. The qualitative study [5] found that physicians' views on accuracy and style were 'largely negative,' especially regarding note length and editing demands. One physician commented that the AI scribe 'writes a novel when a paragraph would do.'
Finally, the evidence is mostly from early adopters and academic settings. The multi-site study [2] had a 60% survey completion rate, which could introduce bias. And the retrospective study [4] cannot prove causation — clinicians who chose to use the AI scribe may have been more motivated or efficient to begin with. So while the direction of the evidence is clear, the magnitude of benefit and the generalizability to all practice settings are still being worked out.
About These Sources
This answer is built on 6 peer-reviewed studies — published from 2024 to 2026, 6 from 2024 or later, 4 in Q1 journals, collectively cited 175 times — selected as the most relevant from 6 studies that passed quality screening, drawn from 38 papers retrieved from a database of over 500 million.
Sources used in this answer
Ambient AI Scribes in Clinical Practice: A Randomized Trial
In a randomized trial of 238 physicians across 14 specialties, Nabla reduced note-writing time by 9.5% versus control, while DAX showed no significant time savings; both platforms improved burnout scores, but occasional inaccuracies were reported.
Use of Ambient AI Scribes to Reduce Administrative Burden and Professional Burnout
In a multi-site quality improvement study of 272 clinicians, burnout dropped from 51.9% to 38.8% after 30 days of AI scribe use, and after-hours documentation time fell by 0.9 hours per day.
Evaluating the Impact of Artificial Intelligence Scribes on Clinical Workflow and Documentation Quality: A Randomized Controlled Trial.
In a randomized trial of 304 orthopedic patients, AI scribes cut physician and administrative time significantly but produced 19 medical errors versus 9 in the conventional group, though all errors were caught before finalization.
Use of an AI Scribe and Electronic Health Record Efficiency.
In a retrospective cohort of 125 AI scribe users, EHR time per appointment fell by 8.5% (2.4 minutes) and time to close encounters dropped by 7.1 hours, with no change in after-hours documentation.
Physician Perspectives on Ambient AI Scribes
In qualitative interviews with 22 physicians, 100% of comments on cognitive demand were positive, but views on accuracy and note length were largely negative, and non-English functionality was limited.
Ambient artificial intelligence scribes: utilization and impact on documentation time
In a prospective pilot with 45 physicians, daily documentation time decreased by 6.89 minutes and total EHR time by 19.95 minutes per day, with notable variation across individual users.