Acceptance and Use of Generative Artificial Intelligence in Higher Education: A UTAUT-Based Model Integrating Trust and Privacy

The rapid emergence of generative artificial intelligence (GAI) is reshaping academic work in higher education. While classical technology acceptance models primarily emphasize cognitive and instrumental determinants, the adoption of GAI also raises ethical concerns related to trust in AI systems and the protection of personal and institutional data. To address this gap, this study examines the determinants of GAI acceptance and use among academic staff in Slovenian higher education institutions by applying a UTAUT-based model that integrates trust and privacy. In this study, GAI is conceptualized as a class of text-based generative AI tools commonly used in academic practice, including applications such as ChatGPT, Copilot, Scholar AI, Gemini, Consensus, and similar systems. A quantitative research design was employed, based on a structured online survey administered to academic staff across 20 higher education institutions in Slovenia (n = 201). Data were analyzed using multilevel confirmatory factor analysis and generalized estimating equations. The results indicate that performance expectancy and attitude toward using significantly predict behavioral intention to use GAI (B = 0.49, p < 0.001 for both), while behavioral intention is the primary predictor of actual use behavior (B = 0.93, p < 0.001). Effort expectancy is positively associated with use behavior independent of behavioral intention (B = 0.23, p = 0.012), whereas trust does not show a statistically significant association with use behavior (B = 0.05, p = 0.458) or behavioral intention (B = −0.01, p = 0.840). Privacy exhibits a positive, but non-statistically significant, association with use behavior (B = 0.12, p = 0.058). The findings highlight the relevance of considering both cognitive and ethical factors when examining generative AI adoption in academic contexts and provide initial empirical insights for refining UTAUT-based frameworks in the context of emerging AI technologies.