Skip to main content

Providence study finds AI ambient listening tool modestly reduces documentation burden, improves provider efficiency

Published Date

RENTON, Wash. [May 29, 2026] — Providence researchers have published a study in JAMA Network Open examining the impact of AI-powered ambient listening technology on provider documentation burden, productivity and efficiency, offering one of the largest real-world evaluations of its kind within a community-based health system. 

The study assessed the use of Microsoft Dragon Copilot, an AI scribe application designed to automatically generate clinical notes from patient-provider conversations. This study builds on previously published research Providence conducted, which focused on a small group of clinicians who self-reported high administrative burden, by broadening the scope to include more than 1,500 physicians and advanced practice providers across multiple specialties. Using electronic health record metadata from July 2023 through March 2025, researchers evaluated objective measures of documentation burden and productivity, including time spent writing notes, after-hours documentation known as pajama time, provider efficiency scores, appointment volume and relative value units. 

“This evaluation allowed us to move beyond perception and really quantify how ambient AI affects provider work,” said Canada Parrish, PhD, principal investigator of the study. “We used objective metadata to measure time in notes, after-hours documentation and efficiency at scale. Even though the individual improvements were modest, when you multiply those gains across thousands of providers in a large community-based health system, the impact becomes meaningful.” Researchers emphasized that even small reductions in uncompensated documentation time can improve provider well-being and job satisfaction. 

The study included 16,149 observation-months from 1,547 active users, defined as providers who used the tool in at least 25 encounters during a given month. Most participants were physicians, and nearly two-thirds practiced in primary care. Researchers found statistically significant reductions in time spent documenting notes during clinic hours following the tool’s adoption, as well as a sustained decrease over time in after-hours documentation. Providers also demonstrated a small but significant increase in relative value units, indicating improved productivity, without an increase in daily appointment volume. 

“These findings align closely with the day-to-day experience of practicing clinicians,” said Scott Smitherman, M.D., associate vice president and chief medical information officer at Providence Clinical Network. “From a physician perspective, ambient AI isn’t about seeing more patients — it’s about reducing cognitive load and reclaiming time spent on administrative work so we can focus more fully on patient care.” 

Lead author Robyn Husa, PhD, reinforced that the timeline of benefits also matters. “We saw a more immediate reduction in time spent in notes, but the decline in after-hours documentation was more gradual. This pattern suggests that the value of using the ambient AI tool may lie more in its sustained use and workflow integration over time rather than immediate benefits across every metric.”  

Dr. Parrish noted that the study’s setting is a key differentiator from much of the existing literature. “Most large-scale research on ambient AI has taken place in academic medical centers,” she said. “As a community-based health system, our results may be more representative and relevant for health systems with similar resources and care models.” 

Overall, the findings contribute to growing evidence that AI ambient listening tools can play a meaningful role in reducing documentation burden and supporting clinical efficiency when deployed thoughtfully at scale.