One AI scribe rollout reached clinicians very unevenly across the same health system
Among 2,901 eligible ambulatory clinicians, adoption ranged from 69.6% in community settings to 34.1% in academic settings. Use does not establish accuracy, safety or benefit.
The short version
- What happened
- Researchers linked electronic-health-record and vendor usage data for 2,901 eligible Johns Hopkins Medicine ambulatory clinicians from December 2024 to May 2025.
- Who it affects
- Clinicians and patients can experience the same health-system technology very differently depending on specialty, workflow and local support; a universal licence does not create uniform access or benefit.
- What remains unproven
- Observational, descriptive design with opt-in use
Keep in mind
Observational, descriptive design with opt-in use
One integrated US health system
Six-month enterprise-rollout window
Unmeasured workflow, consent and local-support factors
Usage logs did not measure note accuracy or editing effort
No safety, clinician well-being or patient outcomes
Source analysis · 1 linked source. Primary analysed source date: 26 October 2026.
Keep following the evidence
Join The Impact Brief founding list for selected AI reports, their effect on people and the evidence limits. Read a sample first.
Read the full report · 3 min The evidence, methods, implications and limitations
Enterprise availability did not produce enterprise-wide use
A peer-reviewed observational study tracked the first six months of an ambient AI scribe's enterprise rollout across Johns Hopkins Medicine. It included every eligible physician or advanced-practice provider with at least one ambulatory visit, producing a denominator of 2,901 clinicians. Electronic-health-record encounters were linked with vendor logs from December 2024 through May 2025; adoption meant using the scribe at least once, while monthly uptake measured the share of eligible visits in which it was used.
The contrast across settings was large. In community practices, 323 of 464 clinicians used the tool at least once—69.6%—compared with 830 of 2,436 academic clinicians, or 34.1%. Community attendings reached 75.0% adoption versus 37.6% among academic attendings; community advanced-practice providers reached 71.8% versus 26.0%. Median visit-level use among community attendings and advanced-practice providers rose from roughly 40% to 80–90%, while many academic groups increased only modestly.[1]
Uptake identifies implementation gaps, not clinical success
Specialty differences were equally striking. Community general paediatrics reached 68.5% adoption and community internal medicine 66.7%, while academic general surgery was 15.6%, obstetrics and gynaecology 17.9% and psychiatry 18.9%. The analysis also described slow-and-steady adopters, early users who stopped, transient users and late adopters. Because participation was opt-in, the patterns may reflect workflow fit, patient consent, local champions, documentation burden or tool performance; the study cannot tell which factor caused them.
Most importantly, a usage log does not show whether an AI-generated note was accurate, safe or efficient to correct. The research did not measure documentation quality, clinician well-being, patient experience or clinical outcomes, and its six-month single-system window cannot establish sustainability or generalisability. Hospitals can use the figures to target workflow assessment and specialty-specific support, but not to claim clinical value. Confidence would rise when adoption data are linked to audited note quality, correction burden, equity, safety and patient outcomes over longer periods. The researchers reported no funding or competing interests.[1]
The distinction that matters
The same rollout produced different adoption patterns
Each percentage uses eligible clinicians in that setting as its denominator and means at least one recorded use.
Community settings
323 of 464 clinicians
69.6% tried the scribe
Academic settings
830 of 2,436 clinicians
34.1% tried the scribe
Clinical benefit
Not measured by usage logs
Accuracy and safety unknown
Watch the background, then check the evidence
Illustrative photo · not a video frame
The Royal Society · 44 min 28 sec · Background explainer
Is this the AI future we want? | The Royal Society
YouTube loads only when you choose; playback starts when you press play.
A Royal Society discussion with Madhumita Murgia and biomedical engineer Alison Noble on AI’s potential in healthcare, science and culture, recorded at the 30 June 2026 event.
Evidence check: A discussion of opportunities and risks, including future possibilities. It is not evidence that every application described is available or clinically proven.
Published 2026-07-01. Source checked 14 October 2026. Check captions on the publisher’s video page.
Check the healthcare evidencePhoto: Daniel Sone / National Cancer Institute / Unsplash · Licence
Continue the story
Read next
Continue exploring patient safety: compare the finding and its evidence limits.
Has Abu Dhabi authorised AI clinicians? No—the new framework is a research pathway
The emirate's health regulator has outlined how systems might eventually qualify to perform defined clinical activities. Physician oversight remains mandatory, while the decisive evidence thresholds and permitted tasks are still to come.
Read the next reportExplore the full source and evidence record
What this means for people
- Clinicians and patients can experience the same health-system technology very differently depending on specialty, workflow and local support; a universal licence does not create uniform access or benefit.
Global context
The health system spans academic hospitals, community hospitals and ambulatory practices in the United States. Vendor configuration, reimbursement, consent rules and documentation workflows differ elsewhere.
What the evidence does not yet show
- Observational, descriptive design with opt-in use
- One integrated US health system
- Six-month enterprise-rollout window
- Unmeasured workflow, consent and local-support factors
- Usage logs did not measure note accuracy or editing effort
- No safety, clinician well-being or patient outcomes
What to watch next
- Audited accuracy and correction burden by specialty
- Patient consent, experience and equity outcomes
- Clinician well-being and after-hours documentation
- Safety events and downstream clinical decisions
- Whether low-use and discontinued groups change after tailored support
Living evidence record
Impact record IAI-0HVJT4P
Evidence stage
Studied
Confidence
Supported
Reporting basis
Source analysis
Independent or research support
Present
Record status
Monitoring
Last checked
26 October 2026
Source trail
1 direct source across 1 source type.
People impact
Documented in this record.
Uncertainty
Limits and next checks are explicit.
Stages describe the evidence available—not whether a technology is good or bad. See the public method.
Single-source reporting disclosure
This record analyses one direct source. It can establish what npj Digital Medicine published or reported, but it is not independent corroboration of every performance claim or predicted outcome. The confidence label will change only when broader evidence is added.
Evidence trail
Sources used for this report
Links checked 26 October 2026
Research question, topics and regions
How clinician adoption and encounter-level use varied during the first six months of a systemwide ambient AI scribe rollout
Topics: Clinical documentation · Health workforce · Implementation · Patient safety
Regions: North America
This report is labelled source analysis. We summarise and analyse source material in our own words; company statements remain attributed claims until independently supported. Translated summaries preserve the meaning of the original source and link back to it. Read our editorial standards.
Reader commentary
What does this mean where you are?
Tell us what you have seen at work, in your community or in everyday life. A question is welcome too. No account is needed; we review notes before publishing them.
Explore commentary across the portal →Published reader notes
0No published reader notes yet. You can start the evidence-led discussion above.
Prefer a private correction or response? Contact the newsroom.