Mid Sweden University

The rapid development of generative AI (GenAI) has affected many areas resulting in discussions on possibilities and challenges. In higher education the main focus has been a focus on assessment and AI resilient assignments where teachers and policy makers have had different opinions. This study has a student perspective, with 16 students examining a Master's program on sustainable information provision. The idea was that students should assess the assessment in the courses that have taken earlier in the program. The aim of the study was both to investigate how sustainable course activities and assignment are, and also to look at the possibilities of integrating GenAI in teaching and learning activities. Moreover, the students have tested Gen AI detection software on AI generated solutions to assignments in the chosen Master's courses.

Data have been collected in group projects where 16 students, divided into 3 groups, investigated 5 courses in each group work. Results were analysed and presented in group reports, combined with individual reflection essays. In the individual essays students were asked to bring up ethical perspectives on GenAI in higher education, and to present suggestion on how the current course design and assessment better should be redesigned for sustainability and fairness. Students carried out this work as a part of a 7.5 ECTS project course in the same Master's programme as the investigated courses belongs to. As input and inspiration to the work in the project course students participated in the Symposium on AI Opportunities and Challenges (SAIOC) in December 2023.

Results show that a majority of the existing assignments in the Master's programme could be partly solved with different GenAI tools. Depending on the nature of the assignments the solutions showed different levels of quality and success. Among many ethical concerns, several student essays brought up the relatively poor quality of the tested detection software, and their doubts of teachers using detection software to evaluate if student submissions are AI generated or not. The recommendation is to provide clear instructions about when GenAI is allowed in course activities, and to redesign for continuous assessment where the whole student journey through a course is assessed and not only on some isolated submissions. Finally, several students bring up the suggestion of oral examinations as a complement to the existing assessment methods.