Informal use of generative artificial intelligence by nursing interns during clinical placement: A multi-stakeholder...

Generative artificial intelligence (AI) has rapidly entered clinical nursing practice. Interns use it on personal devices, yet institutional governance has not kept pace. Existing studies are mostly attitudinal, single-stakeholder, and conducted in academic settings. How interns navigate AI in clinical environments remains poorly understood.

To describe how nursing interns and their supervisors negotiate informal AI use during clinical placement, with implications for nursing curricula and preceptor development.

Interpretive description, with Lipsky's street-level bureaucracy theory as a sensitising framework.

Thirteen tertiary Grade A hospitals in Zhejiang Province, China.

Thirty-two purposively sampled participants across six stakeholder groups: 15 nursing interns, 5 clinical preceptors, 4 head nurses, 5 nursing administrators, 2 information technology specialists, and 1 quality management officer.

Semi-structured face-to-face interviews (40-90 min; February 2026) were analysed through iterative inductive coding and interpretive integration.

Five interconnected themes emerged: (1) multilevel discretionary rule-making in a policy vacuum; (2) a universally endorsed but unevenly enacted verification imperative; (3) intuitive risk stratification; (4) tensions between standardised AI outputs and embodied nursing care; and (5) a cross-stakeholder call for institutional frameworks. Two cross-cutting phenomena further disrupted established hierarchies: a reverse competence flow, in which preceptors recruited digitally proficient interns for AI-mediated tasks; and patient-initiated AI use that eroded the clinician-patient information asymmetry.

This study offers one of the earliest multi-stakeholder accounts of informal AI use in clinical placement. Interns develop sophisticated but unsupported strategies, revealing a new dimension of nursing competence: the capacity to judge when algorithmic knowledge can be trusted, and when embodied clinical judgement must prevail. The findings have direct implications for curriculum design, preceptor development and clinical mentorship. Participants across stakeholder groups converged on the urgent need for tiered governance and AI literacy training for students and preceptors alike.