Designing assessment to maximise student learning in the context of AI
Our assessment strategy is a great place to start. It promotes the idea of ensuring that there is a diversity of assessment across the whole programme to stretch and challenge students to demonstrate their learning in a variety of ways. This prevents students from studying degrees which have a predominance of high-risk assessment types such as certain kinds of essays, where it is harder to verify authorship. It also prevents the predominant use of low risk assessments such as exams, where students do not become digitally literate in generative AI and may rely more on memorisation than creativity and understanding. A key tenet of the assessment strategy is to encourage more authentic assessment which encourages students to develop outputs for a wider audience than the lecturer (such as a podcast). While authentic assessment is not AI-proof, it has the potential to foster greater intrinsic motivation thereby decreasing the likelihood of AI misuse.
Here is a list of examples of assessment that potentially lower the risk of AI misuse:
- Developmental and scaffolded tasks with in-person elements and feedback throughout the term
- Assessments requiring students to describe their process or show their method
- Hand-written components which describe how the final outcome has been achieved (with alternatives for inclusivity.)
- Synoptic assessments which require students to draw on a range of sources and contemporary references
- Public-facing outputs such as podcasts, presentations, posters, and infographics
- Reflective journals showing development over time and making personal connections to learning
- Closed book invigilated in person examinations