When AI meets professional role identity: divergent psychological pathways among physicians and nurses

Purpose This study examines the role-differentiated psychological mechanisms through which perceptions of AI integration shape job satisfaction and turnover intention among hospital physicians and nurses, with particular attention to whether psychological safety functions as a role-contingent mediating mechanism across the two occupational groups. Design/methodology/approach Three-wave longitudinal survey data were collected from physicians and nurses across a large multi-hospital system. Drawing on role theory and the complementary perspectives of uncertainty-bearing and opportunity recognition, the study develops a conceptual model in which occupational role logic moderates the mediating function of psychological safety. Multigroup structural equation modeling was employed to test the proposed pathways and to formally assess cross-group differences in indirect effects. Findings Favorable AI integration perceptions were associated with reduced turnover intention in both groups; however, the pathways to job satisfaction diverged substantially. For physicians, whose role logic centers on bearing professional uncertainty and evaluative responsibility, psychological safety functioned as a critical mediating mechanism – the positive effect of AI integration perceptions on job satisfaction was contingent on experiencing psychological safety. For nurses, whose role logic emphasizes opportunity recognition and task execution, AI integration produced a comparatively direct positive association with job satisfaction, with psychological safety playing a less decisive mediating role. Originality/value This study contributes to research on AI-enabled work and organizational management, by revealing that psychological safety in an AI-enabled work environment functions as a role-contingent mechanism rather than a universal mediator. To explain why this mechanism differs across occupational groups, the study theorizes occupational role logic as a boundary condition that shapes employees' responses to AI integration. Furthermore, it provides findings from a longitudinal, multi-group study, supplementing existing research that has primarily relied on cross-sectional studies of single occupational groups.