Research on the influencing factors of generative artificial intelligence usage intent in post-secondary education: an empirical analysis based on the AIDUA extended model

Objective: Generative Artificial Intelligence (AIGC) presents a profound dialectic in higher education: its transformative potential is challenged by deep-seated psychological and ethical barriers. Traditional adoption models fail to capture this complexity. To bridge this gap, this study develops and tests an integrated cognitive-behavioral framework. We posit that AIGC acceptance is a three-stage cognitive appraisal process. By embedding an extended AIDUA model—a framework specifically tailored to the unique challenges of AI adoption—within Cognitive Appraisal Theory, we investigate how novel antecedent dimensions (Socio-Ethical: ethical risk, explainability; Techno-Performance: generation quality, context-awareness) and classical factors (social influence, hedonic motivation, anthropomorphism) shape core technological beliefs (Performance & Effort Expectancy), which in turn mediate the path to acceptance intention via emotion. Furthermore, the moderating roles of gender, academic background, ethnicity, and political affiliation are systematically examined to test the model’s boundary conditions. Methods: The model was empirically validated using Structural Equation Modeling and multi-group analysis on survey data from 462 university students across 15 diverse institutions in China. Results: The findings reveal that the cognitive appraisal of AIGC is primarily driven by its perceived capabilities and safety. Techno-Performance (generation quality, β = 0.53) and Socio-Ethical (explainability, β = 0.41; ethical risk, β = −0.25) dimensions were the most powerful predictors of Performance Expectancy. These intrinsic appraisals significantly outweighed the influence of external social cues. Notably, ethical risk perception operated as a dual-threat, not only lowering performance expectations but also significantly amplifying the perceived cognitive burden (Effort Expectancy, β = 0.33). Multi-group analyses confirmed that these appraisal pathways are systematically moderated by individual and cultural background variables, highlighting significant heterogeneity in user responses. Discussion: This study makes a critical theoretical contribution by demonstrating how core technological expectancies are formed through a multi-stage appraisal of utility, ethics, and experience, moving beyond mere identification of influential factors. The findings dismantle the myth of a universal “student user,” revealing that AIGC adoption is a culturally and contextually embedded process. Practically, the results provide an evidence-based roadmap for university policymakers and AIGC developers, emphasizing that fostering trust and adoption requires a dual focus: maximizing technological prowess while actively mitigating perceived ethical and cognitive costs through enhanced transparency and user-centric design.