Expectancy and value beliefs predicting generative AI use: evidence from Chinese university faculty

Background: Guided by expectancy-value theory, this study investigated how Chinese university teachers' Generative AI (Gen AI)-specific task values (intrinsic, utility, attainment), perceived Gen AI costs, and Gen AI self-efficacy predict their behavioral intention and frequency of Gen AI use. Methods: From two universities in China, 365 faculty members completed an online survey. Results: Structural equation modeling revealed that utility value and self-efficacy positively predicted behavioral intention, while only self-efficacy was a significant predictor of actual usage frequency. Task values and perceived cost, aside from utility value, did not significantly influence outcomes. Conclusion: These findings highlight the central role of self-efficacy and perceived utility in motivating teachers' adoption of Gen AI, offering theoretical insights for expectancy-value research and practical guidance for professional development initiatives aimed at fostering effective integration of AI in higher education.