A cross-cultural study on the adoption intention of generative artificial intelligence among international students in China

Abstract

Against the backdrop of the deep integration of higher education internationalization and generative artificial intelligence (GAI), international students in China, as a distinctive cross-cultural group, have not been sufficiently explored regarding the internal mechanisms underlying their GAI adoption behavior. This study integrates the Technology Acceptance Model (TAM) and the Unified Theory of Acceptance and Use of Technology (UTAUT2), innovatively introduces the GAI-specific construct “perceived intelligence”, and takes disciplinary background and cultural values (individualism/collectivism) as moderating variables to construct an integrated adoption mechanism model of “external drivers – cognitive/affective evaluation – behavioral intention”. Using stratified multi-stage sampling, 219 international students from 12 universities in eastern, central, and western China were selected as research samples, and partial least squares structural equation modeling (PLS-SEM) was employed for empirical testing. The results show that performance expectancy and perceived intelligence have significant positive effects on perceived usefulness; effort expectancy and facilitating conditions have significant positive effects on perceived ease of use; subjective norms and perceived behavioral control have significant positive effects on attitude; perceived usefulness and attitude significantly and positively drive GAI adoption intention. Perceived usefulness plays a fully mediating role between performance expectancy, perceived intelligence and adoption intention, while attitude plays a fully mediating role between subjective norms, perceived behavioral control and adoption intention. Perceived ease of use has neither a direct impact on adoption intention nor a mediating role. Moderation effect analysis reveals that disciplinary background only has a marginally significant negative moderation effect on the “perceived ease of use→adoption intention” path, and cultural values only have a marginally significant negative moderation effect on the “attitude→adoption intention” path. This study enriches the theoretical system of emerging technology acceptance in cross-cultural educational contexts, provides empirical support for the contextual expansion of classic technology acceptance models, and offers practical references for universities to optimize GAI application strategies in international student education and promote the coordinated development of artificial intelligence and higher education internationalization.