Post #1675460
2026-04-14 12:45 UTC
@liamoc @jfdm @gallais I'm doing something along these lines, using in person discussions to validate student understanding of their code. Specifically, I ask students to journal their debug process: Each symptom they encounter, the steps they took to diagnose it, and the final resolution.
I definitely suspect some GenAI use, but even if I can't attribute it directly, lack of understanding still leads to a worse grade.
That said, in person validation is absolutely not scalable, and only possible because this is the smaller off semester for the class.
Replies (1)
-
@jfdm@discuss.systems 2026-04-14 18:27
@okennedy @liamoc @gallais after speaking to some peers, I’m highly tempted to have ‘class tests’ replace the summative coursework. Only way to ensure no gen ai. Cannot do as fancy, or same indepth work, but it’s a trade off I’m willing to think about to ensure students are tested on fundamentals.