AI scribes cut documentation time, but don't boost productivity

Clinical takeaway: AI ambient scribes may modestly reduce documentation burden, particularly for clinicians who use them frequently, but don't assume they'll improve productivity or eliminate after-hours charting. Ongoing review of note quality and workflow impact remains important.

Ambient AI scribes are being adopted rapidly as health systems search for solutions to EHR-related burnout. However, real-world evidence on whether these tools meaningfully improve clinician efficiency remains limited, and questions persist about their effects on documentation quality, clinician workflows, and patient care.

In a quality improvement study spanning a 14-hospital Boston health system, investigators evaluated the impact of the AI scribe Heidi after unrestricted rollout to outpatient clinicians. The analysis included 902 enrolled clinicians and more than 1.6 million patient encounters. About two-thirds of enrolled clinicians used the tool at least once, while one-third never used it after signing up.

Overall, enrollment was associated with an immediate reduction of about 1.2 minutes in note-related EHR time per encounter. Benefits were concentrated among the heaviest users, who spent about 2.5 fewer minutes per encounter on documentation. For a clinician seeing 10 patients in a clinic session, that translates to roughly 25 minutes saved.

Yet the time savings did not translate into measurable gains in productivity. Work relative value units remained unchanged, and rates of same-day note completion showed little meaningful improvement. Likewise, after-hours documentation did not decrease overall. In fact, heavy users experienced a small increase in after-hours note closures compared with nonusers.

Another notable finding was that notes became longer. Among heavy users, note length increased significantly compared with both low users and nonusers, raising questions about whether AI-generated notes are more complete or simply more verbose.

An accompanying editorial cautioned that apparent efficiency gains may conceal other forms of work. Because the study measured time spent editing notes in the EHR, not within the AI platform itself, it remains unclear whether ambient scribes truly reduce documentation burden or merely shift it elsewhere. The editorial also highlighted unanswered questions about AI-generated omissions, inaccuracies, and how these tools may subtly reshape patients' narratives in the medical record.

As editorialist Kirsten Ostherr, PhD, MPH, wrote, “We must begin asking questions that move beyond quantification to assess the deeper qualitative experience of patients and clinicians with ambient scribes.”

Source: Booth SE, at al. (2026 Oct 27) Ann Intern Med. Impact of Unrestricted Artificial Intelligence Scribe Deployment in a Single Health System on Clinician Workload and Productivity: A Quality Improvement Study