I'm considering generating syntax problem sets in a constructed language, to make AI less useful on assignments. Has anyone already done this or thought about it? I have very mixed feelings about it (and am not thrilled about the amount of work involved).

I’m currently interpreting typos in student papers as a great sign of non-LLM-generated work

we just submitted a manuscript finalizing a project that we've been working on since 2009. Moving at light speed over here.