Understanding the Effect of Risk Perception on the Acceptance and Use of Large Language Models among University Students

Abstract

The rise of Large Language Models (LLMs) in education introduced powerful tools such as ChatGPT, which students increasingly use for academic purposes. However, these technologies present significant challenges for higher education, such as the risk of undermining academic integrity through AI-assisted performance and the uncertainty around proper use. This paper seeks to understand the benefits and risks university students perceive regarding LLM usage and how those influence their acceptance and use of related services in higher education. To this end, we employed a mixed-method approach. Using the UTAUT2 model extended by a Risk Expectancy construct, we conducted an online survey and follow-up interviews with university students. The results indicated that while students perceive considerable risks related to LLMs, those do not impact their usage and behavioral intention. We discuss this phenomenon based on the qualitative interview analysis and suggest research directions for future work.