Academic teachers vs ChatGPT: A comparative study of strategies used by slovenian and polish faculties to prevent dishonest use of generative ai tools

Krzysztof Skonieczny and Valerija Vendramin

Summary:  The article presents the results of a survey of 272 academic teachers in Poland and Slovenia conducted in January and February 2024 concerning prevention of dishonest use of generative AI tools, especially ChatGPT. It focuses on two problems: what teachers understand as dishonest use of such tools; and the strategies they use to prevent such use. No significant differences between the teachers in the two groups were found. The respondents’ answers to the first problem were grouped into three criteria, meaning that the use of ChatGPT can be considered cheating when (1) the AI tool does all the work or part of the work that is deemed significant by the teacher; (2) the use of ChatGPT is not properly disclosed; and (3) the tool is used on certain kinds of assignments, especially exams or theses. In response to the second problem, the respondents agreed that punishing students for using ChatGPT and similar tools is not an effective strategy to prevent dishonest use, while strategies based on human contact, such as discussing the student’s written work with them, evaluating the work in different phases, or explaining what uses of ChatGPT amount to cheating, were deemed more effective.