A study published in JAMA Network Open has found that ambient AI scribes — software that listens to patient visits and drafts clinical notes automatically — were associated with a sharp drop in physician burnout at two major U.S. health systems. Researchers surveyed more than 1,400 clinicians at Mass General Brigham in Massachusetts and Emory Healthcare in Georgia who had used the technology for at least six weeks. At Mass General Brigham, reported burnout prevalence fell from 52.6% before adoption to 30.7% after 84 days of use, a 22-percentage-point decline, with roughly half of that improvement already visible by the 42-day midpoint, when burnout had dropped to 29.4%.
A documentation burden that predates AI
Physician burnout tied to electronic health record documentation has been one of the most persistent complaints in American medicine for over a decade, with studies consistently linking hours spent typing notes after clinic hours — sometimes called "pajama time" — to depression, reduced patient-facing time, and early retirement from clinical practice. Ambient AI scribes attempt to address that burden directly by recording the natural conversation between doctor and patient, transcribing it, and generating a structured draft note that a clinician reviews and edits rather than writes from scratch, cutting the manual typing step out of the workflow entirely.
Emory’s numbers tell a similar story
At Emory Healthcare, researchers tracked a different but complementary measure: the share of clinicians who said documentation technology was having a positive effect on their overall well-being. That figure rose from just 1.6% before clinicians adopted ambient AI tools to 32.3% after 60 days of use — a twentyfold increase that suggests the benefit isn’t limited to one health system’s particular workflow or patient population. Mass General Brigham’s own rollout has grown from an 18-physician pilot in July 2023 to more than 3,000 clinicians with system access by spring 2025, giving researchers a large enough user base to detect statistically meaningful burnout trends rather than anecdotal reports.
Cleveland Clinic’s rollout shows the mechanism at work
A separate, larger deployment at Cleveland Clinic illustrates concretely where the time savings come from. After signing an exclusive five-year partnership with AI scribe vendor Ambience Healthcare in February 2025, Cleveland Clinic onboarded more than 4,000 ambulatory clinicians within roughly sixteen weeks and has since logged over one million AI-assisted patient encounters. Clinicians there report spending about two fewer minutes writing and reviewing notes per appointment, which adds up to roughly 14 fewer minutes of documentation work per physician per day — time that can be redirected either to seeing additional patients or, as many clinicians report preferring, simply going home earlier.
Not a universal fix
Researchers and health-system administrators caution against reading the burnout numbers as proof that ambient AI eliminates documentation burden entirely. Clinicians still must review and correct AI-generated drafts for accuracy, particularly around medication names, dosages, and nuanced clinical judgment calls that a listening algorithm can mishear or misinterpret, and some specialties with highly templated or highly variable visit structures have reported less benefit than primary care and internal medicine. There is also an adjustment period: several clinicians in both the Mass General Brigham and Emory cohorts described early weeks of learning to trust and edit AI drafts as initially adding rather than reducing cognitive load.
Why hospitals are moving fast anyway
Despite those caveats, the adoption curve for AI scribes has accelerated sharply industry-wide, with EHR-integrated AI scribe deployments accounting for roughly 67% of new enterprise documentation contracts in late 2025, up from just 28% in 2022. Health-system leaders point to burnout as a direct driver of costly physician turnover, and the JAMA Network Open data gives hospital administrators one of the first large, peer-reviewed data points connecting a specific AI tool to a specific, measurable drop in a well-validated burnout metric rather than vendor-reported satisfaction surveys.
What comes next
With Mass General Brigham, Emory, and Cleveland Clinic all publishing or presenting real-world results, expect more health systems to publish their own longitudinal burnout data as ambient AI scribes move from early adopters to mainstream infrastructure across ambulatory care. The next open question researchers want answered is durability: whether the burnout gains measured at the 42- and 84-day marks persist a year or more into routine use, or whether some of the initial relief fades as the novelty of the technology wears off and clinicians settle into a new documentation routine.
