The secret of a good assignment could lie in the philosophy of science.
Thinking about doing and doing are two different goals for teaching and training. But thinking in itself is something we do as well, and assignments let the learner train the tools of methodical thinking. If we want to teach students productive working habits when working with AI-tools, we need to teach them to think methodically.
Here are some draft questions:
- The building of Rundetaarn was important to king Christian IV. But what impact could be found in the way of thinking about University Teaching and studying in society?
- What did the participation in building Rundetaarn do to the craftsmen’s understanding of themselves as craftsmen, if anything?
or How did the surrounding town develop financially and culturally because of the building being built? - If Brahe wrote in his journal regarding the building of Rundetaarn, what would he write? How do you interpret that?
- If the goal was to build an observatory for students of the University of Copenhagen with maximum impact, how would you plan, build and test?
- If the goal was to build an observatory for students of the University of Copenhagen with maximum impact, how would you plan, build and test? What would be the tradeoffs for whom with that approach?
- Who build Rundetaarn, and how tall is it?” This does not qualify as an assignment because it is a simple fact that can be googled.
Now add requirements for the hand-in to be well structured, evidence-based with data and arguments aligned with the philosophy of science reflected in the question, critical with an eye for details, including the odd ones out, precise, objective, formal and well-balanced.
These kinds of assignments can IMO be done with the help of AI but require human thinking.