Cheating with AI
A quick little story about students cheating with AI in my recent US History course. Maybe it was a learning experience for them.
Substack is the home for great culture
A quick little story about students cheating with AI in my recent US History course. Maybe it was a learning experience for them.
I said they needed to use at least four of the primary sources we had read in class to answer each question, so I don't think it would be accurate to say I set them up. In a course where we had spent the entire semester reading and discussing these sources, I don't think anyone could reasonably expect that my instructions on the exam could have stretched to include "don't bother engaging with the content as we've been doing since January, but instead ask a chatbot what were a few relevant primary sources dealing with the Revolution or the Civil War."
If a student had done what I asked AND ALSO had said "I found an additional source using AI that I think is relevant", I think that would have been an appropriate use of the tool. I agree that AI can point us to content from which we can learn. But I'm more interested in the students learning than I am in them writing prompts that get the AI to answer exam questions for them. I'm actually a bit surprised you're defending that.
I'm curious what percentage cheated? You said it was "about a half dozen--of how many?" If you made it clear that only the primary sources assigned and discussed could be cited, you're on very solid ground. Even if that was a bit ambiguous, I think you still did the right thing. I do think the Adams letters are an edge case. If no student groups put your head on pike pole, maybe we're seeing the beginning of a thaw in intracampus hostilities.