Perceived Usefulness and Higher-Order Thinking: A PLS-SEM Investigation of the Chaoxing AI Learning Companion Platform among Vocational College Students

Abstract

This study focuses on vocational college students’ adoption of the “embedded” Chaoxing AI learning companion platform and the influence of AI learning companion adoption on higher-order thinking. Unlike general-purpose AI tools, the Chaoxing AI learning companion platform is embedded in institutional teaching platforms and course-based learning processes. However, existing research has largely examined willingness to use general-purpose generative AI tools, and empirical evidence on how AI learning companion functions embedded in educational platforms affect vocational college students’ higher-order thinking remains scarce. Drawing on the Technology Acceptance Model (TAM), this study constructs a research model of “perceived ease of use and perceived usefulness—AI learning companion adoption—higher-order thinking” and conducts PLS-SEM analysis on 75 valid questionnaires using SmartPLS. The results show that the revised measurement model exhibits good reliability and convergent validity. The structural model results indicate that perceived usefulness significantly and positively affects AI learning companion adoption (β = 0.748, P < 0.001), whereas perceived ease of use does not significantly affect AI learning companion adoption (β = 0.092, P = 0.517); AI learning companion adoption significantly and positively affects higher-order thinking (β = 0.540, P < 0.001). The mediation results further show that AI learning companion adoption exerts a significant indirect effect between perceived usefulness and higher-order thinking (β = 0.404, P = 0.001), whereas the indirect effect of perceived ease of use is not significant. The findings suggest that, in the vocational education context, whether students perceive the AI learning companion as “genuinely useful” matters more for their adoption than whether it is “easy to use,” and this adoption is further associated with the development of higher-order thinking. By applying TAM to platform-based AI learning companion scenarios, this study extends the model’s explanatory scope to embedded educational agents and offers practical implications for vocational colleges to optimize AI-supported learning tools and promote more inclusive and sustainable education (SDG 4).