Adoption Intention of Generative Artificial Intelligence Among Chinese College Students: An Extended TAM-UTAUT2 Model from the Four-Helix Perspective

Jiang Xiaomin^1*Kai Cao²Wang Ping³

• ¹Chengdu Polytechnic, Chengdu, China
• ²Qinghai University, Xining, China
• ³Qinghai Nationalities University, Xining, China

Abstract: To explore the formation mechanism of college students’ generative AI (GAI) adoption intention—i.e., dynamic paths of antecedents’ direct/indirect effects and moderators’ interaction effects—this study integrates and extends TAM and UTAUT2. The integration addresses TAM’s overemphasis on rational cognition and UTAUT2’s lack of emotional dimensions and multi-level contextual integration in educational scenarios. A theoretical framework incorporating the “individual-family-institution-region” four-dimensional moderating perspective was constructed, with Partial Least Squares Structural Equation Modeling (PLS-SEM) conducting empirical analysis on 842 college students from 5 universities in eastern and western China (via multi-stage stratified sampling). Questionnaire scales for UTAUT2 core and extended TAM variables were adapted from validated studies. Reliability was tested via Cronbach’s α and Composite Reliability, and validity via Average Variance Extracted, Fornell-Larcker criterion, and Heterotrait-Monotrait Ratio. Results confirmed acceptable reliability/validity, with the structural model showing good explanatory power (R²=0.743). Specifically, perceived comfort (β=0.112, p<0.005), security (β=0.109, p<0.05), and emotional dependence (β=0.497, p<0.005) positively affected perceived usefulness and ease of use. Performance expectancy (β=0.216, p<0.005), social influence (β=-0.064, p<0.05), facilitating conditions (β=0.143, p<0.005), and perceived ease of use (β=0.469, p<0.005) directly drove adoption intention, while perceived usefulness was non-significant (β=-0.031, p=0.523)—challenging TAM’s “rational utility priority” and confirming GAI’s “de-instrumentalization” (reliance on emotional experience and interactive fluency). Moderating effects: Gender negatively moderated performance expectancy-adoption intention (β=-0.207, p<0.05); family structure negatively moderated habit-adoption intention (β=-0.228, p<0.05); university type positively, regional differences negatively moderated performance expectancy-adoption intention (both β=±0.251, p<0.05). Practically, it suggests strengthening western universities’ digital infrastructure, developing differentiated guidance for humanities/science students, and optimizing emotional interaction design. This study provides theoretical support for context-adapted GAI educational application strategies.