Associations between generative AI usage orientations and critical thinking disposition among college students in China

Introduction Research on generative artificial intelligence (GenAI) in higher education commonly describes use in terms of adoption or frequency. Students, however, may engage with the same systems in different ways. Methods This study conceptualises three GenAI usage orientations (Tool, Guidance, and Collaboration) and surveys 430 college students in China to examine their bivariate and adjusted associations with critical thinking disposition (CTD). Tool represented comparatively delegative use, including task completion centred on GenAI and direct adoption of its output with limited independent revision. Guidance reflected instrumental requests for explanations, examples, clues, or diagnostic assistance. Collaboration reflected iterative coordination directed by students. Based on the measurement evaluation, CTD was used as the primary outcome. Results Tool showed no statistically significant bivariate association with CTD ( r = −0.044, p = 0.363) but had a negative adjusted association ( β = −0.189, p < 0.001). This pattern is consistent with statistical suppression; the adjusted Tool coefficient represents a conditional association. Guidance ( β = 0.126, p = 0.017) and Collaboration ( β = 0.321, p < 0.001) showed positive adjusted associations. The orientation block explained an additional 10.37% of the variance in CTD, and the final model explained 38.95% of CTD variance. Discussion The findings show that behavioural orientations provide information beyond general measures of GenAI experience.