As artificial intelligence reshapes the mechanics of clinical practice, its earliest and least-examined casualty may be the cognitive development of the physicians who will care for patients decades from now. The quiet displacement of a foundational training task — writing clinical notes — deserves more scrutiny than it has received in mainstream health discussions.
A six-month pilot at an internal medicine residency program deployed an ambient AI scribe tool across 48 residents, generating roughly 1,000 clinical notes. The authors argue that note-writing is not merely administrative — it is a forcing function for diagnostic synthesis. When trainees construct a clinical note, they are compelled to weigh competing findings, justify therapeutic choices, and build a coherent narrative from incomplete data. Outsourcing that task to an ambient audio-processing system, even a highly accurate one, removes the cognitive struggle that appears essential to developing clinical expertise. The paper proposes seven best practices mapped explicitly to ACGME Core Competencies, including establishing pre-AI documentation baselines, structured scribe training, mandatory critical review of AI-generated text, and redesigned feedback mechanisms.
This work sits at an underappreciated intersection: AI safety not in the technical sense, but in the pedagogical one. The medical education literature has long documented that procedural and cognitive shortcuts during training create durable gaps in expertise — residents who avoid difficult cases develop thinner mental models that surface as errors years later. The concern here is structurally similar. While AI scribes offer genuine efficiency gains for attending physicians managing documentation burden, their wholesale adoption in training environments introduces a substitution risk that competency frameworks were never designed to anticipate. This is a single-institution pilot without a control group or long-term outcomes data, making causal claims premature. Still, as a formative-stage institutional report proposing structured guardrails, it carries meaningful weight — incremental but timely, arriving precisely when residency programs are making adoption decisions that will shape physician reasoning for a generation.