Emergency Physician Perspectives on Impact of Ambient Artificial Intelligence Scribe on Documentation Burden

Human scribes have been shown to reduce physician burnout, while the efficacy of ambient AI scribes—using natural language processing and large language models to generate clinical notes from conversations—is an emerging area of study. This study examines how ambient AI scribe technology affects emergency medicine (EM) resident and attending physician perceptions of documentation burden and burnout.

This retrospective cross-sectional study was conducted at 2 EDs (one urban academic, one community) from September to November 2025. All ED clinicians were eligible to participate, with patient verbal consent for scribe use obtained per institutional policy. Primary endpoints were physician responses to validated survey tools on EMR burden, task load, and self-reported documentation time. Secondary outcomes include post-pilot survey responses to perceived inaccuracies, bias, and patient-safety risk. Survey-based quantitative outcomes were analyzed as absolute pre- to post-intervention changes using multiple t-tests at a 0.05 significance level.

53 EM resident and attending physicians participated in this study. A pre-pilot survey was completed by 40 users (75%); post-pilot survey by 25 (47%). On 5-point Likert scales, perceived time documenting post-shift and EMR burden both decreased (respectively, mean -0.7, 95% CI -1.1 to -0.2, p-value=0.006; mean −0.9, 95% CI -1.3 to -0.4, p-value=0.0004). On-shift task load scores (0-100) declined for rush and effort (respectively, mean −31.7, 95% CI -41.8 to -21.5, p-value <0.0001; mean −20.1, 95% CI -30.8 to -9.4, p-value=0.0004), while cognitive burden and physical demand changes were not significant. No adverse safety events were reported; inaccuracies were rated “occasionally” (mean 3.3) and bias “rarely” (mean 2.0), with hallucinations and excessive detail most commonly cited.

Pilot use of an ambient AI scribe in 2 EDs (academic and community) reduced perceived time documenting post-shift, EMR burden, and task load with only rare to occasional reported inaccuracies and bias. These findings suggest AI scribes may reduce physician burnout in the ED and inform emerging insights around the use of ambient AI tools. This study uniquely explores resident perspectives on documentation burden and the role of ambient AI tools for trainees in the Emergency Department, providing insight to guide future implementation.