Smart and safe use of AI tools for setting assessments

Abstract

Generative AI tools such as ChatGPT, Microsoft Copilot and Gemini have often been promoted in negative ways as they may be used by students in an attempt to breach academic integrity. One issue with assessment question pools is that they are often set and not updated frequently. Hence, they can be overused and are vulnerable to malicious use such as: being sold for money or ported to the dark web and therefore their practical value for assessment is significantly diminished. There is another side of gen AI that can assist academics in authoring assessment questions that is substantially different to the standard question pools that academics often rely on. This paper will illustrate potential uses, benefits, and precautions that need to be taken when using gen AI for assessment design.