AI in higher education: navigating practical benefits and perceived risks from students' perspectives

This quantitative study surveyed 435 university students using a 5-point Likert scale to analyze artificial intelligence (AI) adoption. Grounded in an extended Technology Acceptance Model, the research examines perceived usefulness, control ability, perceived risk, intention to use, and learning impact. Results indicate that perceived usefulness is the strongest predictor of intention to use AI (β = 0.498; p < 0.001), followed by the ability to control and use AI responsibly (β = 0.153; p < 0.001). Conversely, perceived risk has a negative effect (β = −0.189; p < 0.001). Notably, the intention to use AI positively influences learning impact (β = 0.347; p < 0.001) and serves as a partial mediator. The model explains 44.5% of the variance in learning impact. Findings suggest that integrating AI into higher education requires clear pedagogical guidance to maximize benefits while mitigating potential risks.