From How Much to Whodunnit

A framework for authorising and evaluating student AI use

Abstract

The arrival of ChatGPT and other generative AI (genAI) tools has ushered in a new era in education and presented significant challenges to academic institutions. It has also delivered new concerns for educators who seek to support, and to certify, students’ learning. In addition, the potential and in some cases the necessity for students to learn to engage these new tools in preparation for future work in a professional or research context is emerging apace.  This raises important questions for the form and focus of student learning in higher education. It also calls for guidance for educators, especially those who may not be familiar with the operation or implications of these new technologies for their teaching. This paper presents an innovative typology for designing assessment in this context, and that offers language to discuss academic integrity issues and to authorise AI use. The typology draws on and extends scholarship related to groupwork, considering the role of the genAI as a ‘group member’. It provides examples of related approaches to assessment design, and of level descriptors that educators may use as a basis for rubrics to recognise and define the qualities of good student use of genAI tools in this context.