Do AI scribes actually save clinicians time?
Yes, but the savings are modest — roughly 2 minutes per appointment. In a large retrospective study of 125 AI scribe users across an academic health system, clinicians spent 8.5% less time in the electronic health record (EHR) per visit, which translated to about 2.4 minutes saved [1]. The same study found that time spent writing notes dropped by 15.9% (about 1.8 minutes), and the time to close an encounter fell by 7.1 hours [1]. However, there was no significant reduction in after-hours 'pajama time' — the documentation done at home — suggesting the tool didn't fully solve the burnout problem [1].
Physicians themselves report feeling less burdened. In a qualitative study of 22 physicians using an ambient AI scribe, 89% said it reduced their overall workload, and 91% said it improved work-life integration [2]. But these are self-reported perceptions, not objective time logs, and the sample was small and voluntary, so selection bias is possible.
What are the biggest evidence gaps holding AI scribes back?
The most critical gap is the lack of rigorous, replicated trials. A 2022 review noted that almost no AI healthcare tools are independently replicated, meaning a single flawed study could lead to widespread adoption of an unsafe system [4]. The same review found that patient harms are rarely reported in AI trials — a dangerous omission when the tool can hallucinate or omit critical information [4].
Another major gap is accuracy for complex, chronic conditions. A pilot study of 354 primary care visits found that ambient-only AI scribes (listening to the conversation alone) scored only 80.1 out of 100 on documentation completeness for diabetes and hypertension, compared to 94.8 when the tool also had access to the patient's historical EHR data [3]. The biggest difference was in the 'completeness' domain — a 42.5-point gap — meaning the audio-only approach frequently missed important context like medication changes or lab trends [3]. This suggests current tools are not safe for managing chronic disease without deeper EHR integration.
Finally, there is almost no evidence on how these tools perform with non-English-speaking patients, in nursing documentation, or across different specialties. One study found that limited functionality with non-English speakers was a major barrier to adoption [2]. Another paper specifically warns that nurses are being left out of the design and oversight of ambient AI scribes, risking hallucinations, omissions, and bias in nursing notes [5].
Who benefits most from AI scribes, and under what conditions?
Primary care and ambulatory specialty physicians appear to benefit most, especially those who see high patient volumes and spend a lot of time on documentation. In the largest study here, the 8.5% reduction in EHR time was seen across medical subspecialties, surgery, and primary care, but the effect was strongest in primary care [1]. Physicians who already had high baseline documentation burden saw the biggest gains [1].
However, the benefits depend heavily on the tool's design. The qualitative study found that physicians were optimistic about long-term use but frustrated by note length and editing requirements — many felt the AI-generated notes were too long or stylistically off, requiring significant manual correction [2]. Accuracy and style were rated negatively by most physicians [2]. So the tool saves time only if the output is usable enough that editing doesn't eat up the savings.
The evidence also suggests that AI scribes work best when integrated with the patient's full medical history. The pilot study comparing ambient-only vs. history-enhanced documentation showed that the latter was significantly better for chronic disease management [3]. Without that integration, the tool may actually increase risk by missing key information.
About These Sources
This answer is built on 5 peer-reviewed studies — published from 2022 to 2025, 4 from 2024 or later, 1 in Q1 journals — selected as the most relevant from 5 studies that passed quality screening, drawn from 36 papers retrieved from a database of over 500 million.
Sources used in this answer
Use of an AI Scribe and Electronic Health Record Efficiency.
In a retrospective cohort study of 125 AI scribe users and 478 nonusers, AI scribe use was associated with an 8.5% reduction in EHR time per appointment (about 2.4 minutes) and a 15.9% reduction in note time, but no significant change in after-hours documentation or appointment volume.
Physician Perspectives on Ambient AI Scribes.
In a qualitative study of 22 physicians, ambient AI scribes were perceived to reduce cognitive demand (100% of comments), temporal demand (62%), and overall workload (89%), but accuracy and style were rated negatively, and functionality with non-English-speaking patients was a major barrier.
Ambient Only vs. Longitudinal Data-Enhanced AI Documentation: A Pilot Study Quantifying the Value of Historical Clinical Context in Primary Care
In a pilot study of 354 primary care encounters, ambient-only AI documentation scored 80.1/100 on completeness vs. 94.8/100 when augmented with historical EHR data — a 14.6-point difference driven largely by a 42.5-point gap in the completeness domain.
Evidence synthesis, digital scribes, and translational challenges for artificial intelligence in healthcare
A 2022 review identified three persistent translational challenges for AI in healthcare: lack of replication of AI trials, underreporting of patient harms, and performance degradation in different clinical settings.
Invisible Scribes: Can Nurses Trust Ambient AI for Clinical Documentation?
A commentary argues that ambient AI scribes risk hallucinations, omission, and bias in nursing documentation because nurses are excluded from design and oversight, and calls for continuing education and nurse leadership in model development and auditing.