AI Scribe Trims Note Time for Frequent Users, but Broader Workload Gains Stay Elusive

An ambient artificial intelligence scribe program deployed across a large health system reduced the time clinicians spent actively working on patient notes, with the largest savings accruing to those who used the tool most often, according to a quality improvement study published in Annals of Internal Medicine. Researchers from Beth Israel Lahey Health and Beth Israel Deaconess Medical Center analyzed more than 1.6 million patient encounters involving 902 clinicians in the Beth Israel Lahey Health system between July 2024 and January 2026. Their question was twofold: whether enrollment in an AI-powered scribe reduced documentation burden for outpatient clinicians, and whether any benefit varied with how often each clinician actually used the technology. The answers, reported in the study titled Impact of Unrestricted Artificial Intelligence Scribe Deployment in a Single Health System on Clinician Workload and Productivity, were yes on the first count and strongly conditional on the second.

The primary outcome in the study was note-active time per encounter, quantified as the number of minutes a clinician spent with notes open in the electronic health record. This metric captures the period in which a clinician is directly engaged with a note on screen, whether drafting, editing, or reviewing it, and it served as the study’s central measure of documentation workload. After enrollment in the scribe program, clinicians experienced an immediate reduction of 1.2 minutes in note-active time per encounter. That figure represents an average across the full group of 902 clinicians studied, encompassing everyone from intensive users to those who barely engaged with the tool at all. The investigators emphasized that this immediate effect appeared right after enrollment, before any longer-term adaptation to the technology could take hold.

When the researchers stratified clinicians by how frequently they used the scribe, a pronounced gradient emerged in the results. High users experienced a reduction of about 2.5 minutes per encounter in the time spent actively working on notes, roughly double the overall average across the enrolled population. By contrast, minimal change was observed among clinicians who integrated the AI scribe less or not at all into their clinical practice. This pattern indicates that the time savings documented in the study were concentrated among clinicians who made the tool a regular part of their workflow, rather than being distributed evenly across everyone who enrolled. Enrollment alone, in other words, did not guarantee a measurable reduction in note-active time; the magnitude of the benefit tracked the intensity of use.

Alongside the reductions in note-active time, the study examined several related measures of clinician workload and productivity, and on these the results were uniformly negative. Use of the tool did not improve same-day note completion, meaning that clinicians who saved minutes per encounter did not close their notes faster within the day of the visit. It also did not reduce after-hours documentation, the time clinicians spend on notes outside the regular clinic day. Clinical productivity, the third secondary domain the investigators assessed, likewise showed no associated improvement. The published summary states these null findings directly: there were no associated improvements seen in after-hours documentation, same-day note completion, or productivity, despite the measurable reduction in note-active time among enrolled clinicians and especially among frequent users.

The combination of a positive primary outcome and negative secondary outcomes is a defining feature of this study. Clinicians who used the scribe frequently reclaimed about 2.5 minutes per encounter of active note work, yet that reclaimed time did not surface as fewer after-hours documentation minutes, faster same-day completion, or higher productivity as measured in the study. The investigators did not report that the intervention reduced burnout or improved clinician satisfaction, and the published summary makes no such claim. What the study establishes is a specific, quantified reduction in one metric, note-active time per encounter, and the absence of measurable change in three other metrics, within the population, time frame, and health system studied. The authors conclude that ambient AI scribes may help ease documentation workload, especially when used regularly, but that further research is warranted to evaluate their impact on other measures of efficiency.

The scale of the analysis lends weight to these findings. More than 1.6 million patient encounters were included, generated by 902 clinicians across the Beth Israel Lahey Health system over a period running from July 2024 to January 2026. The deployment was unrestricted in the sense reflected in the article title: rather than restricting the scribe to a selected pilot group, the health system made the tool available for clinical use, and the investigators evaluated the resulting range of adoption behavior. This design captured clinicians who embraced the technology and used it heavily, clinicians who used it occasionally, and clinicians who enrolled but integrated it minimally or not at all, allowing the study to relate outcomes to actual usage patterns rather than assuming uniform uptake.

The study also has limitations that frame how its results should be read. It was conducted as a quality improvement study within a single health system, evaluating one deployment context. The study did not examine the quality of the notes produced with the assistance of the AI scribe, a point the authors state explicitly in the published summary. Note quality, including the accuracy of machine-drafted documentation, was therefore outside the scope of the measured outcomes, and the time savings reported in the study pertain to the note-active time metric rather than to any assessment of what the resulting notes contained. The findings apply to the outpatient clinicians, encounters, and time period studied in this health system, and the authors’ own conclusion points toward further research on other measures of efficiency rather than claiming benefits beyond those measured.

The study appears in an issue of Annals of Internal Medicine that also includes an editorial and an Ideas and Opinions piece addressing oversight in clinical artificial intelligence. The Ideas and Opinions article, titled The Illusion of Oversight in Clinical Artificial Intelligence: Human Out of the Loop?, is authored by Christine Y. Zhou, DO; Alexandra Willauer, MD; Eric J. Topol, MD; and Pranav Rajpurkar, PhD. The presence of these companion pieces in the same issue places the scribe deployment study within a broader journal-level conversation about clinical AI, although the published news release does not describe the arguments made in the editorial or the commentary, and their content should be read in the full articles rather than inferred from the deployment findings.

For clinicians and health systems, the study offers a quantified benchmark for what one ambient AI scribe program delivered in routine use at a single large health system. The measured effects were an immediate 1.2-minute reduction in note-active time per encounter across enrolled clinicians, a 2.5-minute reduction among high users, and minimal change among low or non-users, with no associated improvements in after-hours documentation, same-day note completion, or productivity. The authors’ conclusion, as stated in the published summary, is that ambient AI scribes may help ease documentation workload, especially when used regularly, while further research is warranted to evaluate their impact on other measures of efficiency. Corresponding author Sara E. Booth, MPH, of Beth Israel Deaconess Medical Center, is identified in the release as the contact for the study’s authors, and the findings are published under the DOI 10.7326/ANNALS-25-05294 with an article publication date of 27 October 2026.

Subject of Research: Effect of ambient AI medical scribe deployment on clinician documentation time and productivity

Article Title: AI medical scribe cuts clinician documentation time, with greater reduction among more frequent users

Article References: AI medical scribe cuts clinician documentation time, with greater reduction among more frequent users. Original publication (Bibliographic details unavailable.)

Image Credits: AI Generated

DOI: Not provided

Keywords: artificial intelligence, ambient AI scribe, clinical documentation, electronic health record, clinician workload, quality improvement study, Beth Israel Lahey Health, Annals of Internal Medicine, note-active time, productivity, health system deployment, medical