What happened
The Austrian Institute for Health Technology Assessment (AIHTA) published a scoping review of AI for hospital documentation in January 2026. AI scribes record clinician-patient conversations and draft clinical notes with speech recognition and large language models. Seven reviews covered 36 primary studies of them, most from the United States. All seven reviews reported high clinician satisfaction and reductions in perceived documentation burden. Documentation time ranged from a 19-92% decrease to a 13-50% increase. In one randomized trial, scribes drafted patient history sections 2.7 times faster than typing. Productivity measured in relative value units rose by about 4%, with no change in patient volume. Omissions were frequent, and in one trial 36% of notes contained erroneous or fictitious information. No patient safety incidents were reported. AIHTA gives the findings for AI scribes as a group. Most rest on observational studies and on clinicians' own reports.
Sources
-
Assessment report, by authors with no tie to the system.
Erdös J, Grabenhofer L. Artificial Intelligence for Hospital Documentation Support: A Scoping Review of Current Use Cases. AIHTA Project Report No. 171a. Vienna: Austrian Institute for Health Technology Assessment, January 2026. Executive summary and Table 3-3.
Clinician satisfaction with AI scribes was generally high, and documentation burden was perceived as lower, but time savings and productivity gains were inconsistent.
Details
AI system
- Workflow
-
Speech recognition turned clinician-patient conversations into transcripts. Large language models drafted clinical notes for clinicians to review.
Follow-up
- Conditions for replication
-
Clinicians review every draft, since omissions and errors were frequent. Patients agree to have their consultations recorded.
Entry details
- Sector
-
Healthcare (clinical documentation).
- Purpose
-
Several makers, including Abridge, Nabla and Amazon, built AI scribes to turn recorded clinician-patient conversations into draft clinical notes.
Not yet checked: model, release, provider, role, configuration, inputs and outputs, autonomy, human oversight, AI contribution, beneficiaries, reach, distribution and access, duration, every measurement question, every money question, ownership and control, dependencies, cross-border connections, geopolitical implications, replication attempts, availability, institutional decision, grounds, action and result, success criterion, every evidence question, observed or intended, alternative assessments, related entries, other identifiers, category mappings.
Cite this record
Peter H. Vartanian. (2026). Clinicians feel less documentation burden with AI scribes (AIBD-0891). Artificial Intelligence Benefits Database, edition of 31 October 2026. https://petervartanian.xyz/aibd/records/AIBD-0891.html (archived at https://doi.org/10.7910/DVN/AADSII, version 4.0)