AI Literacy and LLM Engagement in Higher Education: A Cross-National Quantitative Study

Abstract

This study presents a cross-national quantitative analysis of how university students in the United States and Bangladesh interact with Large Language Models (LLMs). Based on an online survey of 318 students, results show that LLMs enhance access to information, improve writing, and boost academic performance. However, concerns about overreliance, ethical risks, and critical thinking persist. Guided by the AI Literacy Framework, Expectancy-Value Theory, and Biggs' 3P Model, the study finds that motivational beliefs and technical competencies shape LLM engagement. Significant correlations were found between LLM use and perceived literacy benefits (r = .59, p < .001) and optimism (r = .41, p < .001). ANOVA results showed more frequent use among U.S. students (F = 7.92, p = .005) and STEM majors (F = 18.11, p < .001). Findings support the development of ethical, inclusive, and pedagogically sound frameworks for integrating LLMs in higher education.