Related Experiment Video
Updated: Apr 19, 2026

Augmenting Large Language Models via Vector Embeddings to Improve Domain-Specific Responsiveness
Published on: December 6, 2024
Quality of Clinical Notes Created by Ambient Listening Generative AI: Pragmatic Prospective Pilot Study
Sandra L Taylor1,2, Melissa Jost3, Scott MacDonald3
1Department of Public Health Sciences, School of Medicine, University of California, Davis, 4480 2nd Avenue, Suite 4152, Sacramento, CA, 95817, United States, 1 916-734-4800.
Ambient listening artificial intelligence (AI) scribes reduce physician documentation burden but require careful review. A pilot study found most AI-generated notes had minor errors, but vigilance is needed to prevent patient harm.
Area of Science:
- Medical Informatics
- Clinical Quality Improvement
- Artificial Intelligence in Healthcare
Background:
- Physicians face significant documentation burden from clinic visit notes.
- Ambient listening artificial intelligence (AI) scribes are emerging tools to alleviate this burden.
- Ensuring the safety of AI-generated clinical notes is crucial to prevent potential patient harm.
Purpose of the Study:
- To develop and pilot a standardized, efficient, and scalable method for evaluating AI-generated clinical notes for safety concerns.
- To assess the types and severity of errors in AI-generated notes within ambulatory care settings.
Main Methods:
- A 2-month pilot involving 31 physicians across multiple specialties using an AI scribe for 7545 clinic notes.
- Development of a novel survey instrument to assess note quality, focusing on inclusions, omissions, hallucinations, and bias.
- Physician evaluation of 356 AI-generated notes, rating error severity and analyzing vendor-reported editing metrics.
Main Results:
- Accidental omissions (18%) and hallucinations (11.5%) were the most common errors in AI-generated notes.
- The majority of errors (83.8%) were mild to moderate in severity; only 5.3% posed a serious risk.
- Physician editing varied widely, with a median of 9.0% of AI-generated words changed, and 14.9% of notes left unedited.
Conclusions:
- AI-generated clinical notes are generally high quality, with 94.7% free from significant errors.
- Despite high quality, the potential for serious harm from uncorrected errors necessitates continued clinician review.
- Health systems should pilot AI scribe technology and implement robust review processes to ensure safe adoption.
Related Concept Videos
Non-equilibrium in the Cell
Methods of Documentation I: Source-Oriented Records
In an SOR, each discipline involved in patient care maintains a separate medical record section. This record-keeping method enables easy tracking of patient progress and ensures healthcare staff have access to up-to-date information.
Key Attributes include the following:
Guidelines for Nursing Documentation I
Factual:
The following points emphasize the significance of upholding accurate and unbiased documentation in healthcare.
Purpose of Health Records I
Here's a breakdown of how health records serve these purposes:
Auditory Perception
Techniques of therapeutic communication I: Active Listening, Sharing Observations, Validation, and Using Touch
Therapeutic communication is not the same as social interaction. Social interaction has no goal or purpose and consists of casual information sharing, whereas therapeutic communication has a plan or purpose for the conversation. Therapeutic...