Do AI scribes reduce physician burnout?

The trials so far suggest that AI scribes can reduce burnout scores, but the evidence is encouraging rather than conclusive. The UCLA trial called its burnout results secondary, the scores are self-reported, and the time saved on notes is modest. A scribe may help how clinicians feel about documentation, but no trial has shown that it cures burnout.

The randomized trials point toward lower burnout, measured as a secondary outcome

In the UCLA randomized trial of two AI scribes, the authors concluded that both products resulted in potential improvements in burnout, task load and work exhaustion. They called these secondary findings that need confirmation in larger, multicenter trials ¹.

A University of Wisconsin trial reported a clearer effect on one measure. Ambient AI use significantly reduced work exhaustion and interpersonal disengagement on the Stanford Professional Fulfillment Index, by 0.44 points on a five-point scale ². In the Duke crossover trial, both scribe products reduced personal and work burnout scores, and the difference between the two products on burnout was not meaningful ³.

Taken together, the trials agree on direction. They disagree on size, and each measured burnout alongside other outcomes, such as time on notes.

Burnout is common, so a drop has to be read against the baseline

Burnout among US physicians is widespread, and the level changes over time. In a 2025 Mayo Clinic Proceedings analysis, 45.2 percent of physicians reported at least one symptom of burnout in 2023, compared with 62.8 percent in 2021 ⁴. That national trend has many causes, so a trial's lower score cannot be credited to the scribe alone. Compare each clinic's result with its own starting point and a comparison group.

Time saved on notes explains only part of any change

Burnout is not only about documentation, and the time savings in these trials are small. In the UCLA trial, physicians who used Nabla cut their time-in-note by 9.5 percent against the control group, while users of Microsoft DAX Copilot showed no significant change in time-in-note ^{5 6}. The Wisconsin trial found time on notes fell by 0.36 hours per day, but that was a secondary measure ⁷.

So a clinician can feel less exhausted even when the minutes saved are modest. Documentation is one source of load among several. A scribe can address only that part.

It depends on the product, the measure and the person

• The product. The trials compared named products, and the results differ between them. A score that improved for one product does not prove the same for another.
• The measure. Burnout here is a self-reported symptom screen, not a clinical diagnosis. A trial using a different instrument can reach a different result.
• The design. The trials cited here ran for weeks or months, not years. Whether an effect lasts after the novelty fades remains open.
• The person. Workload, specialty and how much of the visit is conversational all shape how much a scribe helps.

What to do: judge a scribe on the burnout and workload measures you already use

1. Choose a measure you already use, such as a validated burnout survey or your clinic's workload data, before you start.
2. Record a baseline for several weeks before the scribe goes live.
3. Compare with a comparison period or a group that does not use the tool, if your organization can arrange one.
4. Track after-hours time and error reports alongside burnout, since a lower score with more review time may not be a net gain.
5. If you are struggling with burnout yourself, speak with your clinic leadership or occupational health service. A scribe is not a substitute for that support.

This is general information, not medical advice. For the wider subject, see the guide to AI medical scribes. Related questions are how much time doctors spend on documentation and how accurate AI medical scribes are. For the tools themselves, see what an ambient AI scribe is. The overview of dictation for work and study covers the wider subject.