Brian Stone, an associate professor in the Department of Psychological Science, recently published an article in the journal Teaching of Psychology about how students in higher education are navigating generative artificial intelligence (AI). The work was supported in part by the eCampus Center’s Research Fellowship.
Stone surveyed over 700 students at Boise State about their usage of large language models like ChatGPT and Gemini. He found that not only are the vast majority of students familiar with large language models, but over 40% admit to having used AI in ways that are explicitly banned by professors or policy. Indeed, the more familiar students are with generative AI, the more likely they are to have cheated with it. Likewise, students whose professors had integrated or required large language models usage were more likely to have used it in ways not allowed in their courses. “It seems that those using AI the most may not always discriminate between use cases but instead apply the tool broadly,” Stone said.
Student opinion about the new technology was varied, but overall leaned toward nervousness over excitement. Faced with mixed messages from faculty, parents, peers and media, many students find themselves unsure how AI fits into their academic lives and future career. Roughly 60% of students have used AI in circumstances they consider ambiguous (unsure whether such usage is allowed or not). Most students report worrying about false accusations to some extent and more than 10% of students reported having been falsely accused of using AI, with first-generation students possibly more likely than others to report experiencing false accusations. (The most common outcome was being cleared of the accusation).
Stone found that students differed in how relevant they thought generative AI was to their college education or future career, but the vast majority of students endorsed a mercenary, pragmatic view of the workforce: most students believed jobs and industry care more about the product and output of work than they do about the process used to get there.
Students taking online courses were not more likely to cheat with AI than students taking in-person courses. However, there was a significant gender disparity, with men consistently more involved with AI than women. “Given that AI will likely take on an increasingly important role in many areas of the workforce, institutions may want to take steps to ensure students are equitably prepared with AI skills and critical literacy as part of their education,” Stone said. “Indeed, Boise State has recently launched programs like the AI For All certificate to address this need.”