Abstract
Abstract
Background
To identify joint latent profiles of AI literacy and critical thinking among clinical nurses and to examine the differential associations of these two competencies with evidence-informed decision-making (EDM) competence across profiles.
Methods
A multicenter cross-sectional survey was conducted from March to June 2026 among 2,306 registered nurses recruited from eight hospitals across four Chinese provinces using stratified convenience sampling. Data were collected using the AI Literacy Scale, the Clinical Nurses' Critical Thinking Assessment Scale, and the EDM Competence Scale. Latent profile analysis was used to identify joint profiles, subgroup-specific regression to examine the associations of the two competencies with EDM across profiles, and multinomial logistic regression to identify factors associated with profile membership, all adjusted for covariates.
Results
Four profiles were identified: low literacy–low critical thinking (22.38%), low literacy–high critical thinking (35.78%), high literacy–moderate critical thinking (20.56%), and high literacy–high critical thinking (21.29%). EDM competence followed a clear gradient: high–high > high–moderate > low–high > low–low (P P
< 0.001). Subgroup-specific regression revealed that in the high–high group, both AI literacy (β = 0.432) and critical thinking (β = 0.351) demonstrated strong independent effects on EDM, whereas in the high–moderate group, the effect size of AI literacy (β = 0.296) was markedly attenuated. AI-specific training (OR = 1.813) and routine AI use (OR = 1.465) positively predicted high–high membership; notably, routine AI use showed the strongest predictive effect for the high–moderate profile (OR = 1.991).
Conclusion
Clinical nurses exhibit four heterogeneous joint profiles of AI literacy and critical thinking. The high literacy–moderate critical thinking group (20.56%) represents a hidden-risk population—proficient in technical operation but deficient in critical appraisal, in whom AI use behaviors may exert a double-edged sword effect. This stratified framework can inform targeted training, competency-based job allocation, and quality control of AI-assisted decision-making.