Adapting existing assessment

The following five approaches offer practical options for lecturers who wish to reduce opportunities in assesment to outsource core thinking to AI tools. They can be used individually or in combination, depending on the course, discipline, and learning outcomes.

1. Convert the assessment to a supervised format

Replacing an unsupervised assignment with a supervised format ensures that students demonstrate their understanding in real time, without access to AI tools.

Example: replace a take-home essay with an on-site exam or oral assessment. Find more supervised formats.

2. Add a verification step

Verification steps help lecturers confirm that the submitted work reflects the student’s own understanding of the material. If assignments are prepared outside class, it may be necessary to verify that the work reflects the student’s own understanding. Clearly communicating the purpose of verification steps can help students understand how these activities support the learning objectives of the course.

Possible verification points include:

Example: Keep the written assignment but add a brief oral defence or one or two exam questions that ask students to explain and justify key choices based on their submitted work.

3. Make the learning process visible

Rather than evaluating only the final product, focus on how ideas develop, how sources are selected and used, and how decisions are made throughout the process. This approach also strengthens formative assessment and feedback, which has a strong positive effect on student learning (Hattie & Timperley, 2007; Boud & Molloy, 2013).

Feedback does not have to mean more work for you. Students learn most effectively from a combination of self‑assessment, peer feedback, and instructor feedback (Wiliam, 2013), so not all feedback needs to come from the lecturer.

Example: Ask students to submit a research proposal, one or more drafts, an annotated bibliography, or a short methodological explanation alongside their final work, so that the development of their thinking becomes visible.

4. Redesign the task

Design tasks that require disciplinary knowledge, specialised methods or frameworks, and active engagement with course‑specific materials. This can include:

• interpreting datasets, primary sources, or case studies
• analysing unseen materials during the assessment
• justifying methodological or analytical choices

Using recent, unique, or course‑specific materials makes it harder for AI tools to generate adequate responses without genuine disciplinary understanding, and makes it more difficult for students to outsource their thinking.

Examples accross disciplines

Where students are likely to use AI for non‑essential aspects of written work — such as language polishing, basic structure, or summaries — these elements can carry less weight in grading. You can reduce their weighting or assess them on a pass/fail basis. Rubrics can then place greater emphasis on higher‑order skills such as analysis, argumentation, interpretation, and originality.

Example: Reduce the weighting for “language and style” from 30% to 10%, and increase the weighting for “analysis and argumentation” from 40% to 60%, while keeping other criteria (such as use of sources) unchanged.

Research projects and theses are central components of many degree programmes. Because they involve large amounts of unsupervised work, they require particular attention when designing assessment in the context of generative AI.

It is therefore crucial to verify the students’ understanding and ownership of their work by incorporating supervised verification moments. In many programmes, research projects already include multiple stages, which naturally lend themselves as such verification points, for example:

• research proposals or project plans
• discussions of literature reviews or theoretical frameworks
• presentations of research design or methodology
• draft submissions and feedback meetings
• supervision discussions during data analysis

These checkpoints help supervisors observe how the student’s thinking develops throughout the project. Many programmes also include an oral defense or viva, where students explain their research decisions and respond to questions about their work. This provides an additional opportunity to verify that the student understands and can justify the research.