From usefulness to motivation: a TAM–SDT perspective on AI adoption and problematic internet use in higher education

The rapid adoption of generative artificial intelligence (AI) tools such as ChatGPT in higher education necessitates a clearer understanding of their influence on technology acceptance and learning motivation. Grounded in the Technology Acceptance Model (TAM) and Self-Determination Theory (SDT), this study examines the relationships among perceived usefulness (PU), perceived ease of use (PEOU), ChatGPT use (GPTU), problematic internet use (PIU), and learning motivation (LM). A cross-sectional survey was conducted among university students, and the data were analyzed using Partial Least Squares Structural Equation Modeling (PLS-SEM). Reliability and validity were confirmed through composite reliability, Cronbach’s alpha, and average variance extracted (AVE). The structural model revealed that PU significantly predicts GPTU ( β = 0.419, p < 0.001), whereas PEOU demonstrates a weaker and non-significant direct effect ( β = 0.220, p = 0.089). GPTU positively influences LM ( β = 0.323, p < 0.001), while PIU shows a negative effect on LM. The model explains a moderate proportion of variance in GPTU and LM, and mediation analysis indicates that PU indirectly enhances learning motivation through increased ChatGPT use. These findings underscore the central role of perceived usefulness in generative AI adoption and highlight the dual impact of AI tools in educational contexts, where beneficial learning outcomes may coexist with risks of excessive or problematic use. The study contributes theoretically by integrating TAM and SDT in a generative AI context and offers practical implications for responsible AI implementation in higher education.