AI literacy as a potential mediator between attitude and self-efficacy among PICU nurses: a cross-sectional study

Background Artificial intelligence (AI) is reshaping the healthcare landscape, particularly in high-stakes settings like pediatric intensive care units (PICU). While AI holds potential to enhance clinical care, its integration into nursing practice may depend on nurses’ attitudes, literacy, and self-efficacy regarding AI. However, the relationships among these constructs—and whether AI literacy plays an indirect role in the attitude–self-efficacy association—remain underexamined in PICU nurses. Aim This cross-sectional study examined whether AI literacy is statistically consistent with a mediating role in the relationship between PICU nurses’ AI attitude and AI self-efficacy. Design A multicenter cross-sectional study conducted in Sichuan Province, China. Methods A convenience sample of 221 registered nurses from 10 PICUs in one Chinese province completed self-report measures of the General Attitudes toward Artificial Intelligence Scale (GAAIS), the Artificial Intelligence Literacy Scale (AILS), and the Artificial Intelligence Self-Efficacy Scale (AISES). Data analyses included Pearson correlation and mediation analysis (PROCESS Model 4) with 5,000 bootstrap samples. Given the cross-sectional design, all analyses are associational, not causal. Results PICU nurses reported generally positive AI attitude, AI literacy, and AI self-efficacy on the respective scales. Positive correlations were observed among all three variables (all p < 0.01), with the strongest association observed between AI literacy and self-efficacy ( r = 0.743). The indirect effect of AI attitude on AI self-efficacy via AI literacy was statistically consistent with mediation [indirect effect = 0.815, 95% CI (0.582, 1.082)], accounting for 76% of the total effect. Regarding subgroup differences, nurses with advanced degrees and those employed in teaching hospitals showed higher mean scores in AI literacy and self-efficacy ( p < 0.05). Conclusion In this cross-sectional sample of PICU nurses, the findings were consistent with an indirect role of AI literacy in the association between AI attitude and AI self-efficacy. However, because the data are cross-sectional, causality cannot be inferred. These results are hypothesis-generating and support future longitudinal and intervention studies to test whether enhancing AI literacy can improve AI self-efficacy over time. Pending such confirmation, nursing administrators may consider integrating AI literacy into continuing education as one potential strategy to support nurses’ AI self-efficacy.