Student Use of Large Language Model Artificial Intelligence on a History of Mathematics Module (CPD Case Study)
Prof Isobel Falconer (School of Maths and Statistics)
[email protected]
- What motivated you to use AI in this module, and what goals or challenges were you aiming to address?
A pragmatic acceptance that we would not be able to detect reliably whether students were using AI in their project or not, plus a belief that use of AI was going to be part of most students’ future employment, and we should encourage them to think critically about how and when to use it.
2. How did you design or adapt the assessment, and how did you prepare students for using AI appropriately?
We kept the former assessment – an extended essay of a history of maths topic of the student’s choice – but asked also for two compulsory paragraphs:
- A paragraph evaluating the ways in which the student had used/not used LLM AI and explaining their decisions. The intention was to be even-handed by asking that all students justify their decisions, whether or not they decided to use LLM AI;
- A paragraph identifying the three most significant sources cited in their essay and what these had contributed to their argument. This paragraph acted as a check that they did, indeed, understand both the argument they had presented, and were familiar with at least some of their sources
We also changed the marking criteria from previous years to put more weight on quality of argument (which AI is poor at), up to 50% from 40%, and less weight on presentation (which AI is good at), down to 10% from 20%.
We discussed the pros and cons of AI use in a tutorial
3. What challenges did you encounter, and how did you address them?
We encountered remarkably few challenges. A couple of students were very unhappy that we were allowing AI but accepted it and seemed satisfied in the end that grading was fair.
The other challenge was the additional marking time assessing the additional paragraphs against the essay. But this was trivial compared with the time to mark the essay thoroughly anyway, and probably no greater than the additional time it would have taken to read essays carefully for signs of AI.
4. What benefits did you see for students and for your own teaching practice?
Students gained critical awareness, through experience, of what AI could, and could not, do for them.
Unfortunately, as I’m not lecturing any more, remarkably little!
5. How did you evaluate the usefulness of this assessment to ensure that it reflected the desirable learning outcomes?
Aspects of AI use were not an explicit learning outcome for this assessment. However, analysis of both quantitative data (grades obtained) and qualitative data (the students’ paragraphs on their use of AI) provided much information for evaluation – this is discussed in my following publication. I published reflections on our experience, and you’ll also find the relevant question rubrics etc in the appendices:
Falconer, Isobel, ‘Student Use of Large Language Model Artificial Intelligence on a History of Mathematics Module’, MSOR Connections, 23.2 (2025), art. 2 <https://journals.gre.ac.uk/index.php/msor/article/view/1540>
6. What would you do differently next time, and what advice would you give to colleagues?
If I were doing it again, I would give rather more tutorial time to preparation for AI use/non-use, perhaps in a more structured way than just discussion. And, if there is any chance at all that one might want to publish outcomes, then get ethics clearance in advance (we didn’t, and it hampered what I could do with the rich information gained)