Dong Zhang, Guijun Fu, Kai Cao, X Q Wang, Huihui Li
In the baseline UTAUT model, both performance expectancy (PE) and effort expectancy (EE) significantly predicted CBI. In the full model (integrating SCK, KCS and PT), only PE remained a significant direct effect. KCS significantly influenced PT, which in turn predicted CBI; SCK showed no such effect. The full model explained 56.3% of the variance in CBI.
Background Generative artificial intelligence is reshaping K-12 education, yet teachers' adoption remains unbalanced and under-researched. This study integrates TAM, UTAUT, and Perceived Trust Theory to construct an adoption model for K-12 teachers' GAI adoption intention. Methods A survey of 443 K-12 teachers from 16 prefecture-level cities in Shandong Province was conducted. PLS-SEM with bootstrapping (5,000 resamples) was employed to test direct, mediating, and moderating effects. Results Performance expectancy, perceived trust, and facilitating conditions significantly predicted perceived ease of use. Effort expectancy showed a marginal positive effect on ease of use but a significant negative effect on usefulness. Perceived ease of use and usefulness positively predicted attitude, which in turn drove adoption intention ( β = 0.678, p < 0.001). A chain mediating path of “perceived ease of use → perceived usefulness → attitude” was confirmed. Moderating effects of demographic characteristics were path-specific and limited. Conclusion Attitude serves as the critical hub connecting rational cognition to behavioral intention. Effort expectancy exhibits a dual effect specific to K-12 educational contexts. Differentiated support strategies are needed for teachers of varying backgrounds.