AI Literacy and Perceived Reliability as Predictors of Student Trust in Generative AI-Assisted Decision Making

Generative artificial intelligence is increasingly used by university students to compare alternatives, obtain recommendations, and support academic decisions. This study examines whether objective AI literacy and perceived reliability can predict student trust in generative AI-assisted decision-making. A cross-sectional online survey yielded 406 responses. After eligibility screening, 310 respondents met the inclusion criteria, and 280 respondents who passed the attention check served as the primary analysis sample. AI literacy was measured using 12 multiple-choice questions, while perceived reliability and user trust were measured on a five-point scale. Item analysis supported the removal of two negatively worded items that had low item-total correlations. Multiple linear regression with HC3 heteroskedasticity-robust standard errors showed that the model explained 52.1% of the variance in user trust. Perceived reliability was the dominant predictor (standardized beta = 0.704; p < 0.001), whereas AI literacy provided a smaller positive contribution (standardized beta = 0.101; p = 0.022). The AI literacy coefficient was non-significant in a sensitivity analysis that retained all eligible respondents. These findings indicate that student trust primarily depends on the assessment that generative AI outputs are accurate, consistent, relevant, and reliable. Higher education institutions need to integrate AI literacy education with verification practices so that students are capable of assessing system reliability before using AI recommendations.