Keeping AI in check: how clinicians govern AI-generated knowledge in hearing care

Purpose This study aims to examine how clinicians use AI-enabled tools and govern the knowledge generated by these systems in clinical decision-making. Specifically, it explores how clinicians identify and manage risks, uncertainties and responsibilities associated with AI-generated outputs before using them in patient-facing decisions. Design/methodology/approach The study adopted a qualitative research design, collecting clinicians’ narratives through open-ended essays. Data were analyzed using the Gioia methodology with an inductive-abductive approach grounded in the knowledge governance perspective. Findings The study develops a knowledge governance pathway to help clinicians manage AI-generated knowledge in practice. The pathway includes surfacing epistemic vulnerabilities, subjecting AI-generated outputs to evidentiary scrutiny, allocating decision authority between AI and clinicians and institutionalizing traceability and adaptive oversight. The findings show that clinicians treat AI outputs as provisional, context-dependent knowledge claims and rely on triangulation, verification routines and professional judgment before using them in patient-facing decisions. They also sustain effective AI use through documentation, transparency, training and continuous recalibration. Originality/value This study advances the understanding of AI use in health care by shifting attention from ethical and technical concerns to knowledge governance in clinical practice. By providing a clinician-centered process model grounded in clinicians’ lived experiences, it offers actionable insights for health-care professionals and organizations seeking to integrate AI into clinical decision-making while preserving professional judgment and accountability. It also extends knowledge governance theory by showing how clinicians validate AI-generated outputs, allocate decision authority between AI and clinicians and sustain accountability through traceable oversight before using these outputs in patient-facing decisions.