Session 6: Designing assessments in higher education: ChatGPT and academic integrity

Abstract

The 2023/24 academic year was marked by the new, widespread availability of open-access AI tools. This technology offers exciting opportunities for streamlining tedious tasks in personal, research, and teaching settings, as well as assisting in improving the structuring and writing skills of students. However, it also raises deep concerns among educators in Higher Education about the challenge of swiftly and comprehensively updating assessments to prevent cheating. Rapidly changing large language models, highly variable individual technological skills among both teachers and students, and insecurity regarding general guidance for “AI proofing” made it difficult to predict with confidence how susceptible assignment questions and assessment types would be this year.

This TLA-funded project investigates the performance of ChatGPT4 across more than 100 assessments from 40 participating modules in the School of Biological Sciences.

By comparing the grades achieved by the ChatBot with those of the student cohort, we aim to assess the extent to which AI tools may be used as “cheating aid” across different assessments, identify assignment types that are inherently resistant to AI cheating, and uncover subject-specific and universally applicable best practice in assessment design. The modules in the School of Biological and Environmental Sciences offer a wide range of often highly creative assessments, but these have not yet been investigated in a comparative manner. I expect to identify innovative, currently hidden, examples of best practice that extend beyond mere AI-proofing, offering insights for innovative assessment strategies across disciplines.

Designing assessments in higher education: ChatGPT and academic integrity PowerPoint. Only LJMU staff and students have access to this resource.