University students’ perceptions and adoptions of AI: a cross-national study

Abstract

Recent advances in Artificial Intelligence (AI), especially large language models (LLMs), are rapidly transforming educational practices worldwide. Despite growing adoption, there is limited cross-national evidence on how university students perceive and engage with these technologies. This study addresses this gap by analyzing data from 1906 students across 13 countries, using a novel psychometric approach that leverages LLMs for item generation and validation. Our findings reveal that students’ trust in AI, familiarity with algorithms, and lower anxiety about AI are key predictors of positive perceptions and adoption of LLMs in learning. Notably, most students with even minimal trust in AI reported using LLMs, primarily for coding, idea generation, and writing, and perceived them as at least moderately effective. In contrast, students who did not use LLMs were unlikely to use other AI tools. These results underscore the importance of fostering trust and AI literacy to support effective integration of AI technologies in tertiary education. Practical implications include the need for targeted training programs, transparent data practices, and strategies to address student anxiety and promote responsible use. By providing cross-national insights and introducing innovative measurement tools, this study offers actionable recommendations for educators, institutions, and policymakers seeking to enhance student engagement with AI in diverse educational contexts.