Rethinking "explain and demonstrate"

[StevenClontz]

Right now, we're doing much better to scaffold our assessments than the old days where I'd get "yes" or "no" to the question "do these vectors span $\mathbb R^5$?".

But I'm starting to get a little frustrated with students who are memorizing or copying incantations, causing responses that I'd expect from an LLM rather than a human who is writing words they understand. I wonder if one way to address this is to introduce more variety to the prompts where we ask students for their explanations, particularly later in the semester when there's a variety of possible arguments. If the prompt they get is random, then they need to actually read the prompt of course, and while they can still have a more complicated algorithm for matching prompts to canned responses, I'm hopeful this would push at least some students into improving the thoughtfulness of their responses.

This is particularly imporant in the situation where I am now: I have last-minute submissions I'm reviewing, and I don't have the luxury to send it back to them for clarifications, so I'd prefer to have more obvoius tells for when a student isn't being thoughtful in their writing vs. has a crucial misunderstanding.