Artificial intelligence-induced attentional displacement in clinical diagnosis

PREPRINT — This manuscript has not undergone peer review. A version of this manuscript is currently under consideration by the Journal of the American Medical Informatics Association (JAMIA). Objective: To describe artificial intelligence-induced attentional displacement (AIIAD) as a proposed mechanism affecting clinical diagnosis. Perspective: AI-mediated selection, summarisation, ranking, and framing may shift attention away from patient history, examination, and discordant findings, even without explicit acceptance of an AI recommendation. A published patient-facing chatbot case illustrates the concern but does not establish the mechanism in clinicians. AIIAD overlaps with automation bias, attentional tunneling, misplaced salience, anchoring, and premature closure; its independent explanatory value remains unproven. Implications: Evaluation should examine the combined clinician-system workflow and changes in attention, information seeking, and diagnostic reconsideration. Source visibility, preservation of high-consequence clues, independent differential formation, and diagnostic pauses warrant prospective evaluation. AI should extend clinical attention rather than appropriate it.