The use of AI large language models 1 by university students for assignment preparation.

Abstract

Using an opportunity where students were explicitly permitted to use artificial intelligence (AI) applications to prepare an assignment, we compared the practices and beliefs of two distinct student cohorts: second-year Science students at a large metropolitan university in Australia and students at a partner institution in the People's Republic of China. Data from an anonymous survey revealed that over 50% of students employed AI tools, primarily for idea generation and initial drafting, instead of using these tools as a source of text they would use in the assignments without editing or attribution. Reasons for use showed differences between native English speakers, their non-native speaking classmates, and their classmates attending the PRC campus. Across the cohorts, there was a measured engagement with AI tools when preparing assignments. Notably, all cohorts declared a greater willingness to use such tools if institutional licenses were made available. Most students believed that they would use AI in their future professional roles and that their institutions should be providing guidance on the proper use of AI tools. At the Australian campus, sanctioning the use of AI in preparing the class assignment was not associated with any change in the average assignment marks from those gained by a cohort from the previous year where no sanction was in place. At the same time, there was a significant improvement in the average mark for the PRC students, which was associated with the same sanction.