Randomised trials published in 2025 and 2026 found that ambient AI scribes reduce the time doctors spend on notes and improve measures of burnout, with effects that differ between products. The same trials report occasional inaccuracies, so the doctor has to read and sign every note.
What an ambient scribe is
An ambient scribe listens to the consultation, with the patient's agreement, and drafts the clinical note from the conversation. The doctor reviews the draft, corrects it and signs it. Some scribes also draft referral letters and patient instructions from the same conversation.
The evidence changed in the past year. Until late 2025 no randomised trial of ambient scribes had been published, according to the authors of the UCLA trial.1 Three randomised trials now report results, alongside earlier before and after studies such as one across six US health systems, in which the share of clinicians reporting burnout fell from 51.9% to 38.8% after 30 days with a scribe.2 Two of its authors were employed by an ambient scribe company, which the paper discloses.
The earlier study is useful context. It surveyed 263 ambulatory clinicians before and 30 days after they started using a scribe, and also found less time documenting after hours, more focused attention on patients, and a better ability to add urgent patients to the schedule.2 It was a quality improvement study based on self-reports, so it shows association, and the randomised trials are the stronger test.
The UCLA trial
At UCLA Health, 238 outpatient physicians in 14 specialties were randomly assigned to one of two scribes or to usual care between November 2024 and January 2025, in a trial published in NEJM AI.1 One scribe cut time in notes by 9.5% against the control group, and the other showed no significant change.1 Users of both reported improvements in burnout, task load and work exhaustion scores, which the authors say need confirming in larger trials.1
The Wisconsin trial
At the University of Wisconsin, 66 practitioners took part in a 24 week randomised trial across ambulatory clinics in two states.3 Of 71,487 notes written, 38% were drafted with ambient AI. Use was linked to a significant fall in work exhaustion and interpersonal disengagement, and to about 22 minutes less time on notes each day, with no loss of note quality, diagnosis coding or billing compliance.3 Professional fulfilment did not rise significantly.3
The Duke trial
At Duke, a randomised crossover trial of 160 outpatient clinicians compared two scribes. Both reduced personal and work burnout scores, one saved about three more minutes a day in notes than the other, and after-hours time in the record was largely unchanged.4
What the trials agree on
Time on notes falls by minutes a day. Burnout and task load scores improve. Results differ between products, so a clinic should test a scribe on its own consultations before it commits. And clinicians in the UCLA trial reported inaccuracies that the doctor has to catch.1
Products differ
Two of the three trials compared products directly. At UCLA, one scribe reduced time in notes and the other did not.1 At Duke, clinicians reported a larger improvement in workflow satisfaction with one product than the other, 2.51 against 1.91 points on a seven point scale, and the same product saved more time in notes.4 A scribe that works well in one clinic's specialties, languages and note templates may not in another's.
What the trials did not show
The Wisconsin trial found no significant rise in professional fulfilment, and its reduction in work outside work hours lost significance once extreme values were removed.3 At Duke, time spent in the record outside scheduled hours was largely unaffected.4 The UCLA authors describe their burnout findings as needing confirmation in larger, multicentre trials.1 Trials so far ran for weeks or months in US outpatient settings.
Accuracy and the signature
Clinicians in the UCLA trial said clinically significant inaccuracies occurred occasionally, and one mild adverse event was reported.1 The authors concluded that occasional inaccuracies require ongoing vigilance.1 Dubai's policy on AI in healthcare describes supportive technology as a tool that assists a human who makes the final decision and is responsible for it.5 A scribe that sends nothing until the doctor signs keeps that responsibility where the policy puts it.
Rules for a Dubai clinic
The DHA's AI policy applies to every DHA licensed facility that uses AI.5 It asks clinics to report incidents involving AI tools to the DHA, to make sure every user understands a tool's strengths and limitations, to educate patients about the AI in use, and to arrange regular independent audits.5
The recording itself is health data. Federal Law No. 2 of 2019 requires health data to be kept confidential and bars its use for non-health purposes without the patient's written consent or another listed exception.6 Ask the vendor how long recordings are kept, where they are hosted and who can hear them, and get each answer in writing. wai hosts your clinic's records in the UAE, under the UAE's rules. We set this up with your clinic.
The policy also expects AI tools to fail safely, with automatic alerts and a gradual stop if hardware or software malfunctions.5 Ask the vendor to show what the doctor sees when the scribe cannot hear the room.
How to trial a scribe
- Pick two or three doctors in different specialties and run the scribe for four weeks.
- Measure time in notes and after-hours time in the record before and during the trial.
- Ask each doctor to log every correction they make, by type.
- Tell patients in Arabic and English that an AI scribe is drafting the note, and ask for their agreement.
- Keep the scribe if the doctors want to keep it and the correction log is falling.
Where wai fits
wai Clinicians writes the note while the doctor and patient talk, drafts the letter before the patient leaves, and sends nothing until the doctor has signed it. The doctor finishes clinic with the notes done.
Questions
Do AI scribes save doctors time?
In randomised trials, ambient scribes reduced time on notes, by about 22 minutes a day in the University of Wisconsin trial. The effect differed between products, so trial a scribe on your own consultations.
Are AI scribes accurate?
Clinicians in a UCLA randomised trial reported clinically significant inaccuracies occasionally. The doctor should read and sign every note before it is filed or sent.
Do AI scribes reduce burnout?
Trials at UCLA, Wisconsin and Duke all found improvements in burnout or work exhaustion scores. The authors call for larger studies to confirm the size of the effect.
Can a clinic in Dubai use an AI scribe?
Yes, within the DHA's policy on AI in healthcare, which expects clinician supervision, incident reporting and patient education. Federal Law No. 2 of 2019 on health data also applies to the recordings. wai hosts your clinic's records in the UAE, under the UAE's rules. We set this up with your clinic.
Sources
- 1Lukac PJ and colleagues, 2025Ambient AI Scribes in Clinical Practice, a Randomized Trial, NEJM AIRead source
- 2Olson KD and colleagues, 2025Use of Ambient AI Scribes to Reduce Administrative Burden and Professional Burnout, JAMA Network OpenRead source
- 3Afshar M and colleagues, 2025A Pragmatic Randomized Controlled Trial of Ambient Artificial Intelligence to Improve Health Practitioner Well-Being, NEJM AIRead source
- 4Chowdhury A and colleagues, 2026Comparing ambient scribes, a randomized crossover clinical trial addressing ambient scribe technologies' impact on physician burnout, Journal of the American Medical Informatics AssociationRead source
- 5Dubai Health Authority, 2026Artificial Intelligence in the Healthcare policy, DHA/HISHD/PP-09, effective 2021, revision dateRead source
- 6Latham and Watkins, 2019New UAE Law Regulates Healthcare Data, Client Alert 2514Read source
Written and reviewed by the clinical team at wai. Last reviewed 29 September 2026.
