Evaluation of Socio-Technical Mechanisms Shaping AI Scribe Documentation Failures: A Netnographic Study.

BACKGROUND

Artificial Intelligence (AI) scribes are increasingly adopted to address electronic health record (EHR) documentation burden. Although early evaluations report perceived efficiency gains and reduced after-hours work, findings on documentation quality and safety remain mixed. Reported issues, including omissions, attribution mistakes, and hallucinated content, raise concerns about potential clinical, administrative and medico-legal risks. Existing evaluations largely focus on performance metrics, offering limited insight into the socio-technical conditions shaping real-world experiences and outcomes.

AIM: To examine how interacting socio-technical conditions influence AI scribe use, reported problems and associated risk implications in clinical documentation.

METHODOLOGY: A netnographic analysis was conducted of 2267 relevant data segments from 952 documents across 162 Reddit threads (2023-2025) drawn from clinician-oriented communities. Data were collected using a structured query design via the Python Reddit API Wrapper. Data were analysed using an inductive-abductive qualitative approach and organised through a socio-technical lens across technology, organisation, person and environment domains.

RESULTS

Contributors' accounts suggested that reported AI scribe problems were associated with interacting technological constraints, including template rigidity, integration gaps and reliability issues; organisational governance and billing pressures; environmental time constraints; and individual verification practices. These conditions appeared to operate through three mediating mechanisms: adoption and configuration practices, workflow coupling and documentation targets. Reported problems included content-related issues, such as misattribution, hallucinations and omissions, as well as workflow disruptions, including latency, crashes and copy-and-paste friction. Clinicians described potential clinical, administrative and medico-legal risks as contingent on integration quality, governance clarity and review capacity.

CONCLUSION

AI scribe safety is not solely a function of model accuracy. The findings suggest that documentation problems may arise through socio-technical interactions that influence whether errors are identified, corrected or carried forward. Safe deployment requires strengthening integration, governance and verification processes alongside technical performance.