A prospective study conducted at Hawaii Pacific Health Medical Group has found that an ambient AI documentation tool meaningfully reduced physician burnout symptoms and after-hours charting, adding fresh evidence to the debate over whether AI scribes can genuinely fix one of medicine’s most persistent workplace problems. The research, published in JMIR Medical Informatics in 2026, tracked clinicians who piloted an AI note-writing system called Abridge, integrated directly into the practice’s Epic electronic health record, from initial rollout through several months of real-world use.
Inside the Pilot
The study recruited 80 providers across the medical group, with 79 completing the full pilot. Over the course of the trial, those clinicians generated more than 25,000 AI-assisted notes spanning 23 different medical specialties, giving researchers an unusually broad look at how ambient AI performs across everything from primary care to specialty clinics rather than in a single department. Participants were grouped by how often they actually used the AI scribe during patient encounters, with “high use” defined as reliance on the tool in 60% or more of visits, allowing the researchers to compare outcomes between enthusiastic adopters and more occasional users.
Notable Drops in After-Hours Work
The results point to a substantial reduction in the phenomenon clinicians call “pajama time,” the hours spent finishing medical notes at home after clinic hours. High users of the AI scribe saw a 21% decrease in total time spent on notes per day, amounting to roughly 13.6 fewer minutes daily, along with a 13% drop in pajama time specifically, about 3.6 minutes less per day. Perhaps more strikingly, before the pilot, 25 of the 79 providers, or 31.6%, reported spending eight or more hours per week on notes outside of clinic hours. After adopting the AI scribe, that number fell to just 6 providers, or 7.6%, a 76% relative decrease in the share of clinicians doing heavy after-hours documentation work.
Measurable Gains in Burnout Symptoms
Burnout, which researchers typically measure using validated self-report scales covering emotional exhaustion and depersonalization, also improved. With the ambient AI tool in use, 39 of the 79 providers, or 49.4%, reported no burnout symptoms at all, a 22% relative increase compared to baseline. The study’s authors note that provider-perceived workload decreased alongside these burnout improvements, while self-reported note quality remained favorable, addressing a common worry that offloading documentation to AI might come at the cost of accuracy or thoroughness in the medical record.
Part of a Broader Wave of AI Scribe Research
The Hawaii Pacific Health findings arrive amid a broader surge of research into ambient AI documentation tools across U.S. health systems in 2025 and 2026. Separate research from UChicago Medicine found that clinicians using ambient AI spent 8.5% less total time in the electronic health record than matched colleagues who did not use the tool, along with a more than 15% reduction in time spent specifically composing notes. Multiple randomized controlled trials are now underway comparing different ambient AI scribe products head-to-head, including studies focused on emergency medicine practitioners and pediatric subspecialists, reflecting how quickly health systems have moved from small pilots to systematic evaluation of these tools.
Why Proponents Say This Matters Beyond Convenience
Physician burnout has been linked in prior research to worse patient safety outcomes, higher rates of medical errors, and significant physician attrition, making documentation burden a issue health system leaders treat as a strategic priority rather than a minor administrative annoyance. Supporters of ambient AI scribes argue that results like those from Hawaii Pacific Health show these tools can meaningfully restore time and energy to clinicians without requiring them to change how they interact with patients, since the AI listens to and transcribes the natural conversation rather than requiring structured data entry during the visit itself.
Remaining Questions About Scale and Durability
Skeptics caution that pilot studies conducted at a single medical group, even one spanning 23 specialties, may not generalize cleanly to larger or more resource-constrained health systems, particularly safety-net hospitals or rural practices with different patient populations and staffing levels. There are also open questions about whether burnout improvements measured after a few months persist over a longer horizon, once the novelty of a new tool wears off, and whether note quality holds up under the kind of independent chart audits that go beyond clinicians’ own self-assessments. As more randomized trials comparing competing AI scribe products report results over the next year, health systems will have a clearer picture of whether findings like those from Hawaii Pacific Health represent a durable fix for physician burnout or an early, optimistic snapshot of a technology still being stress-tested at scale.
Photo: Engin_Akyurt / PIXABAY via Pixabay
