The relationship between AI self-efficacy and critical thinking among nursing students: A network analysis.

AIM: This study used network analysis to characterize the internal structure of artificial intelligence self-efficacy (AISE) among nursing students and to examine how specific AISE components related to critical thinking (CT). BACKGROUND: As artificial intelligence (AI) becomes embedded in nursing education, individual differences in AISE may shape students' engagement with AI-supported learning. Concerns have emerged that efficiency-oriented AI use may alter cognitive engagement, with implications for CT. DESIGN: Descriptive cross-sectional study. METHODS: A total of 611 Chinese nursing students completed the Artificial Intelligence Self-Efficacy Scale (AISES) and Critical Thinking Scale (CTS). Network analysis was conducted in R using LASSO-regularized partial correlation networks, with node centrality, bridge strength and predictability estimated via bootstrapping. RESULTS: The AISE network comprised 22 nodes and 121 edges (52.38% of possible edges; mean weight = 0.093). AI_4 ("AI tone matches humans") showed the highest strength centrality, whereas AI_1 ("AI interaction process is vivid") emerged as the strongest bridge node. The integrated AISE-CT network included 28 nodes and 160 non-zero edges (42.33% of possible edges; mean weight = 0.088). AS_7 ("AI easy to control") was the primary bridge linking AISE and CT. Anthropomorphic interaction and time-saving features showed small, opposite associations with Maturity. Truth-Seeking showed the highest predictability among CT dimensions (R² = 0.850). CONCLUSIONS: Network analysis identified anthropomorphic interaction as the central AISE component and perceived controllability as the key bridge between AISE and CT. AISE components showed heterogeneous associations with CT dimensions, emphasizing critical appraisal and reflective engagement beyond perceived efficiency alone.