Ambient AI Scribes as Emerging Infrastructure in the Learning Health System

Area of Science:

• Health Informatics
• Artificial Intelligence in Medicine
• Health Systems Research

Background:

• Ambient AI scribes automate clinical documentation from clinician-patient conversations, rapidly deploying in US health systems.
• Initial evaluations suggest benefits like reduced clinician burden and improved well-being, creating a perception of inevitability.
• However, outsourcing the initial documentation process significantly alters clinical data production and the infrastructure of learning health systems (LHSs).

Purpose of the Study:

• To analyze ambient AI scribes not just as workflow tools but as critical emerging infrastructure for LHSs.
• To conceptualize clinical documentation as the epistemic substrate for generating analyzable data powering healthcare analytics and institutional learning.
• To identify and address potential risks associated with algorithmically mediated documentation in LHSs.

Main Methods:

• Drawing on infrastructure studies and LHS frameworks to analyze the role of clinical documentation.
• Synthesizing emerging evidence on the risks and impacts of ambient AI scribes.
• Proposing a governance agenda grounded in LHS principles for oversight.

Main Results:

• Ambient AI scribes, while offering efficiency, introduce risks of systematic documentation errors (e.g., hallucinations, omissions) due to proprietary systems and data.
• These errors can propagate invisibly through analytic pipelines, affecting quality measurement, predictive modeling, and clinical decision support.
• Differential performance across diverse patient populations (accents, speech patterns) poses equity concerns.

Conclusions:

• Ambient AI scribes function as infrastructure, materially shaping LHS capabilities and potentially eroding data integrity and equity.
• Risks include altered documentation norms and data quality before downstream effects are assessed.
• A governance agenda focusing on quality metrics, drift monitoring, equity evaluation, transparency, and multi-stakeholder stewardship is essential for trustworthy LHSs.