Generative AI dependency and perceived learning outcomes among university students: an indirect association involving effort regulation

Objective Generative artificial intelligence (GenAI) is increasingly integrated into university learning, raising questions about dependency-oriented reliance and self-regulation. Drawing on a self-regulated learning perspective, this study examined associations among GenAI dependency, effort regulation, and multidimensional perceived learning outcomes, including the statistical indirect association involving effort regulation. Methods Cross-sectional survey data were collected from 982 students at three higher education institutions in Guangdong Province, China. Confirmatory factor analysis and covariance-based structural equation modeling were conducted, with indirect associations tested using 5,000 bias-corrected bootstrap resamples. Sensitivity analyses adjusted for GenAI use frequency. Results The measurement model showed acceptable fit (CFI = 0.953, TLI = 0.944, RMSEA = 0.043, SRMR = 0.041). GenAI dependency showed a small negative total association with perceived learning outcomes ( β = −0.087) and was negatively associated with effort regulation ( β = −0.257). Effort regulation was positively associated with perceived learning outcomes ( β = 0.497). The residual direct association was nonsignificant ( β = 0.041), whereas the indirect association was negative and significant ( β = −0.128). After adjustment for GenAI use frequency, the indirect association remained significant, but the small total association was no longer significant. Conclusion The findings highlight effort regulation as a relevant learner-controlled process in the association between GenAI dependency and perceived learning outcomes. They also distinguish dependency-oriented reliance from frequency of GenAI use. The cross-sectional design and potential classroom clustering warrant cautious interpretation.