Exploring the intention-behaviour gap in Artificial Intelligence adoption: the case study of students’ ChatGPT usage in Dutch higher education

This paper investigates the AI intention–behaviour gap in higher education, defined as the misalignment between students' stated intention to use generative AI for study-related purposes and their enacted use. Although behavioural intention is central in technology acceptance research, prior evidence indicates that intention explains only a modest share of variance in technology use, suggesting that intention and behaviour may diverge. Building on this premise, I examine how AI-specific perceptions and individual characteristics predict (a) students' intention to use ChatGPT and (b) their self-reported usage behaviour, allowing a direct test of whether determinants of intention and behaviour overlap or differ. Survey data were collected from 310 higher-education students in the Netherlands. Results indicate a substantive intention–behaviour gap, with intention to use ChatGPT being more prevalent than actual use. Logistic regression analyses show that emotional creepiness, need for human interaction, and self-efficacy are associated with both intention and use, whereas trust predicts intention but not use, and observability predicts use but not intention. These findings advance research on human-AI interaction and educational technology adoption by demonstrating that predictors of intention and enacted use are not fully aligned for generative AI tools in higher education. Practically, the results provide support for the notion that efforts to support effective implementation should not focus solely on increasing positive attitudes, but also on reducing affective barriers, strengthening students’ capability to use generative AI responsibly, and improving the visibility of acceptable use practices within institutions.