The Verdict on Ambient AI Scribes: Real Time Savings, But Only for Consistent Users, JAMA Study Finds

Health systems have spent heavily on ambient AI documentation on the promise of fixing clinician burnout. The strongest controlled evidence yet on whether it works arrived in JAMA this spring, from UCSF and Mass General Brigham researchers in the Ambient Clinical Documentation Collaborative — and the answer is: yes, but with important fine print.

The study tracked 8,581 ambulatory clinicians across five academic health systems between June 2023 and August 2025, comparing 1,809 adopters of ambient scribes against 6,772 non-adopters. The tools studied — Ambience, Nuance DAX Copilot, and Abridge, all Epic-integrated — delivered 13 fewer minutes of total EHR time and 16 fewer minutes of documentation time per 8-hour clinical day.

The gains concentrated where the documentation burden is heaviest: primary care clinicians saved around 25 minutes a day, and advanced practice clinicians roughly 40. Adopters also squeezed in about half an additional visit per week and roughly $167 more in monthly billing revenue per clinician.

The fine print

Two findings complicate the ROI story. First, only 32% of users hit the 50%-of-visits usage threshold associated with maximum benefit — meaning most clinicians used the tools too inconsistently to capture the full gains. Second, after-hours "pajama time," the burnout metric health systems most want to fix, did not decrease significantly overall.

For buyers, the implication is clear: the technology works, but the return depends on adoption discipline, specialty mix, and workflow integration — not the license count.

Reporting: Telehealth.org, on the JAMA multisite study published April 1, 2026.