AI literacy and AI anxiety in nursing students: the serial mediating roles of attitudes and self-efficacy
TL;DR
Higher AI literacy was associated with lower AI anxiety, and this association was partly accounted for by AI attitudes and AI self-efficacy in the proposed serial mediation model, which suggests that more favorable attitudes may be linked to stronger self-efficacy, which may be related to lower anxiety.
Abstract
Background The rapid integration of artificial intelligence (AI) into health care demands that future nurses develop foundational AI competencies. However, the mechanisms through which AI literacy alleviates AI-related anxiety among nursing students remain inadequately understood, particularly regarding the sequential psychological processes involved. Objective This study aimed to test whether attitudes toward AI and AI self-efficacy mediate the relationship between AI literacy and AI anxiety in nursing students, and to examine the serial mediation pathway among these variables. Methods A multicenter cross-sectional survey was conducted among 1,482 nursing students from 11 universities in Sichuan Province, China, between May and October 2025. Validated Chinese versions of the Artificial Intelligence Literacy Scale (AILS), General Attitudes toward Artificial Intelligence Scale (GAAIS), Artificial Intelligence Self-Efficacy Scale (AISES), and Artificial Intelligence Anxiety Scale (AIAS) were administered. Structural equation modeling (SEM) was employed to examine direct and indirect effects. Results AI literacy was positively correlated with favorable AI attitudes (r = 0.285) and AI self-efficacy (r = 0.509), and negatively correlated with AI anxiety (r = −0.434). The SEM showed a significant negative direct association between AI literacy and AI anxiety (β = −0.329, P < 0.001). Significant indirect associations were observed through AI attitudes and AI self-efficacy separately and through the serial pathway AI literacy → attitudes → self-efficacy → anxiety. These findings describe model-based associations and should not be interpreted as evidence of temporal or causal ordering. Conclusion Higher AI literacy was associated with lower AI anxiety, and this association was partly accounted for by AI attitudes and AI self-efficacy in the proposed serial mediation model. The observed ordering suggests that more favorable attitudes may be linked to stronger self-efficacy, which may in turn be related to lower anxiety; however, the cross-sectional design does not establish causality. Nursing education may therefore consider combining AI knowledge and skills training with opportunities to develop balanced attitudes and confidence in using AI.