Perceptions and trust of AI among health care professionals: A mixed methods study before and after a pilot interruption

Abstract

Artificial intelligence (AI) is increasingly introduced into health care to improve efficiency, support decision‑making, and reduce administrative burden. However, many AI projects stall before full implementation, and little is known about how such interruptions shape professionals’ perceptions or trust. This study investigates health care professionals’ perceptions of AI before implementation and examines how an unexpected pilot interruption influenced their views on organisational readiness, technological maturity, and institutional trust. An explanatory sequential mixed‑methods design was used. A pre‑pilot survey (N = 75) captured baseline attitudes toward AI, understood here as structured evaluative perceptions. Quantitative analyses included descriptive statistics, group comparisons, and correlation analysis. Semi‑structured thematic interviews (n = 6) followed and were analysed using theory‑driven qualitative content analysis informed by sensemaking theory. Survey results showed generally positive initial attitudes, especially among younger professionals and those with previous AI experience. Higher perceived AI competence strongly correlated with more favourable attitudes. Interview findings indicated that the pilot interruption did not substantially change technology‑related attitudes. Instead, it triggered reinterpretations of organizational readiness, resource allocation, and the maturity of the technology. The interruption was interpreted as affecting institutional trust more than trust in the AI system itself, with participants linking the event to communication practices, timing, and broader organisational challenges. The study suggests that interrupted implementations function as critical sensemaking events in which institutional trust becomes the primary object of re-evaluation, whereas attitudes toward AI remain comparatively stable. The findings highlight the importance of transparent communication, realistic expectations, and organisational preparedness when managing AI adoption.