Can Large Language Models Write to the Curriculum?

The question

I am preparing a paper on a specific question: can large language models produce mathematics assessment items for cycle 3 that comply with the Plan d’études romand (PER), the curriculum of French-speaking Switzerland?

Why it matters

Generating a set of exercises now takes a few seconds. But a correct exercise is not necessarily aligned with the objective set by the PER, nor pitched at the right level of demand for the class. The aim is not to replace the teacher who designs their assessments, but to find out what these tools are really worth when they are held to the requirements of the curriculum.

Three dimensions

1. Alignment: does the item match the PER objective it is meant to assess?
2. Cognitive demand: what level of thinking does the item ask of the pupil?
3. Item quality: is the item correct, clear and usable in class?

Status and materials

Study in preparation, carried out alongside my master’s in education research (UNIR). The study protocol, materials and preprint will be added to the OSF project osf.io/b653h.

• assessment
• cognitive demand
• curriculum alignment
• generative AI
• item generation
• large language models
• mathematics education
• Plan d’études romand
• Switzerland

Further reading

What AI really changes for teachers who prepare and assess — the article that sets out the classroom stakes behind this research.

Working on these questions in French-speaking Switzerland? My contact details are on the home page.