Exploring AI in Higher Education: A Structural Equation Model of Perception, Usage, and Learning Effectiveness

This study investigates how undergraduate students’ perceptions of AI tools influence their actual usage patterns and subsequent learning outcomes, using structural equation modeling to analyze data from online and blended learning courses across two universities.

College students are using tools like ChatGPT and other AI programs to help with studying, writing, and organizing their learning. This project looks at how students feel about these tools—whether they find them useful, easy to use, or concerning—and how those feelings affect how they actually use the tools. We then see whether using AI tools helps them become better students or feel more satisfied with their learning. The results help teachers and schools figure out how to best use AI in classes.

Perceived Usefulness and Enhanced Learning Experience are the strongest predictors of AI tool engagement and satisfaction.

Concern (e.g., about ethics or privacy) was not significantly related to usage or outcomes in the model.

Accessibility (ease of use and availability) plays an important mediating role in connecting positive perceptions to learning outcomes.

Students who report higher academic skills also report greater reliance on AI tools, more interactive learning, and increased satisfaction.

A surprising finding: Interactive Learning (AI-mediated engagement) had a negative relationship with overall learning outcomes, suggesting a possible overreliance or superficial engagement.