The association between large language model dependence and academic anxiety among nursing students: statistical indirect associations through artificial intelligence addiction and academic burnout

Background Increasing use of large language models in nursing education has raised questions about the association between reliance on these tools and academic distress. This study examined direct and specific indirect associations between large language model dependence (LMD) and academic anxiety (AA) through artificial intelligence addiction (AIA) and academic burnout (AB). Methods A cross-sectional survey was conducted from March to April 2026 among 473 nursing students in Guangdong Province, China, including students undertaking clinical placements. The four constructs were assessed using items selected from published self-report instruments. Partial least squares structural equation modeling was performed with age and sex included as covariates. Direct and specific indirect associations were evaluated using bootstrapping with 5,000 resamples. Results Cronbach’s α ranged from 0.914 to 0.944, composite reliability from 0.935 to 0.954, and average variance extracted from 0.560 to 0.721. Heterotrait–monotrait ratios ranged from 0.485 to 0.733, supporting discriminant validity. Model fit evidence was mixed: SRMR was 0.068, whereas NFI was 0.812, below the 0.90 reference value. The model explained 49.2% of the variance in academic anxiety. Large language model dependence showed a small positive direct association with academic anxiety ( β = 0.138, 95% CI [0.006, 0.278], p = 0.048). Significant specific indirect associations were identified through academic burnout ( β = 0.205, 95% CI [0.123, 0.303]) and through artificial intelligence addiction followed by academic burnout ( β = 0.098, 95% CI [0.049, 0.147]); both p < 0.001. Conclusion Large language model dependence was positively associated with academic anxiety among nursing students, with specific indirect associations through academic burnout and through artificial intelligence addiction followed by academic burnout. These findings highlight the relevance of problematic AI use and study-related exhaustion to academic distress. However, the cross-sectional design does not establish temporal ordering or causal mediation.