Panuwat Sikham/istock via Getty

News

Study: AI scribes boost physician productivity, increase revenue

A recent study revealed that physicians are gaining productivity benefits from AI scribes without a detrimental effect on claim denials.

By
  • Brian T. Horowitz
Published: 26 Jan 2026

Ambient AI brings real increases in physician productivity and revenue, according to a study published Jan. 9  in JAMA Network Open.

The study revealed a 5.8% increase in relative value units (RVUs) compared with non-adopters. RVUs are a measure of value that CMS uses to decide physician payment.

The per-week increase of 1.81 RVUs seen in the study enables health systems to offset expenses they have incurred when using AI scribes, translating to $3,044 in additional annual revenue per physician, according to University of California, San Francisco researchers. The authors arrived at that dollar amount using the 2025 Medicare Physician Fee Schedule.

Meanwhile, AI scribe adopters were able to handle 0.80 additional patient encounters each week, comprising a 2.8% boost in visit volume, according to the study.

"AI scribe adoption was associated with increases in RVUs and encounters per week, with no evidence of increased denials," the authors stated.

The lack of difference in the proportion of denied claims between adopters and non-adopters indicates that AI-generated documentation adhered to payer standards, the study revealed.

"These findings provide initial evidence that AI scribes change how physicians code and bill, and may increase productivity and patient access to care," lead author A Jay Holmgren, Ph.D., associate chief for research in the UCSF Division of Clinical Informatics and Digital Transformation (DoC-IT), said in a news release. Other UCSF researchers for the study included UCSF physician-hospitalist Cynthia L. Fenton, MD, and Robert Thombley, lead data architect for the Center for Clinical Informatics and Improvement Research at UCSF.

Examples of AI scribes include Microsoft Dragon Copilot and Suki, which capture a full patient-provider conversation and integrate it into a clinician's workflow.  

How UCSF researchers conducted the study

For the study, UCSF researchers analyzed close to 1.2 million ambulatory encounters among 1,565 physicians between January 2023 and April 2025. Among the physicians who participated, 698, or 44.6%, used AI scribes, per UCSF.

UCSF researchers pulled EHR metadata at UCSF Health during this time frame for ambulatory visits by attending physicians.

Researchers conducted the study according to a set of reporting guidelines called STROBE, which stands for the Strengthening of the Reporting of Observational Studies in Epidemiology. Its checklist helps authors write analytical observational studies.

Fueling future research on AI scribes

The UCSF study focused on a single health system, which limited the scope of the results, according to the authors. In addition, voluntary early adopters participated, which may deliver different data than typical physicians regarding health outcomes. The researchers attributed this discrepancy to possible bias.

"Study limitations include EHR metadata from one site, which may not capture the universe of confounders, and potential self-selection bias in early adopters," the study stated.

Going forward, the researchers plan to evaluate factors in how health systems use AI scribes and determine whether increased RVUs mean additional clinical services or accurate coding rather than upcoding, the authors wrote. Upcoding is a type of medical fraud in which providers inflate codes.

Researchers will also "strengthen causal association using randomization or other natural experiments."

Brian T. Horowitz started covering health IT news in 2010 and the tech beat overall in 1996.

Dig Deeper on Clinical documentation