<p>Despite the growing integration of AI in universities, the influence of cultural dimensions on students’ behavioural intention and actual use remains underexplored, particularly in Sub-Saharan Africa. This study investigates the predictive role of Hofstede’s cultural dimensions in shaping university students’ behavioural intention and actual use of AI within a culturally adapted Unified Technology Acceptance Model (UTAUT) framework in higher education. A cross-sectional, correlational, quantitative study was conducted, with 102 students participating by completing an online questionnaire. Structural equation modelling results indicated that personal innovativeness significantly influenced behavioural intention. Individualism-collectivism, behavioural intention, personal innovativeness, and social influence had significant effects on behavioural use, whereas indulgence-constraint, power distance, and uncertainty avoidance had no effect on either behavioural intention or behavioural use. Cultural dimensions did not significantly influence Social Influence in this sample; however, their effects may still be meaningful in other educational contexts. Predictive performance was assessed using PLS-Predict (10-fold, 10-repeat) and the cross-validated predictive ability test (CVPAT). Results indicate that while the PLS-SEM model outperformed the naïve indicator average (IA) benchmark, it exhibited weaker forecasting ability than the linear regression benchmarks. These findings highlight the importance of complementing explanatory models with predictive assessments when evaluating technology adoption in higher education contexts. In addition, the findings will inform lecturers on how to craft interventions grounded in personal innovativeness, social influence, and collectivism (Ubuntu) to foster AI use, as well as inform policymakers and higher education institutions that cultural alignment (framing) alone is insufficient to drive AI adoption.</p>