Generative AI is a promising adjunct to blended learning, offering an innovative means to enhance academic performance. Its rapid diffusion has been accompanied by criticism and uncertainty, particularly regarding ethics and the potential displacement of human labor. A review of the existing research reveals persistent gaps in understanding AI use among students. This study therefore aimed to develop an integrated model to explain generative AI adoption across two distinctive time points. Employing a survey-based design, cross-sectional data were collected at two time points from college students at a local tertiary institution in Hong Kong. PLS-SEM Model testing showed that performance expectancy was the strongest and most persistent determinant of both intention to use and actual use across both data collections. Risk propensity had no effect at the outset, but at a longer usage time point, it was significantly related to intention and use through performance expectancy. Social influence exerted a direct and significant effect initially and later demonstrated both direct and indirect significant effects on intention and use via performance expectancy. The findings identify key determinants and enhance our understanding of the complex decision-making process involved in the use of generative AI.
Will W. K. Ma (2026) studied this question.