There are a variety if "guides" out there to try to provide some sort of structure and sense to the question, "Should a student use AI on this assignment?" None of them are very useful.
Let's take this example:
That "Generative AI Acceptable Use Scale" has been run in EdWeek and used by at least one actual instructor. It was adapted by Vera Cubero (North Carolina Department of Public Instruction) and based on the work of Dr. Leon Furze, Dr. Mike Perkins, Dr. Jasper Roe FHEA, and Dr. Jason Mcvaugh. That's a lot of doctors. And yet.
The disclosure requirements are cute, in that way that classroom teachers recognize not so much as "I'm sure you will follow these requirements" so much as "I'm going to express my expectations clearly so that later, when you try to ignore them, there will already be a foundation for my complaints about what you've done."
But lets expand on the guidelines themselves. Because in my rural area, I can envision a student who lives without enough reliable wifi to connect to ChatGPT, but happens to live next door to a smart graduate student-- let's call that grad student "Bob."*
So with the AI guide in mind, let's craft some rules for Acceptable Use of Bob for assignments.
No Bob Use, also known as the Don't Cheat option, is of course the preferred default.
Bob-Assisted Idea Generation and Structuring. In this option, Bob would come up with the idea for your paper, and/or provide you with an outline for your work. The continued acceptance among AI-mongers that idea-generation and structuring are not really part of the writing process, and therefor it's okay to have Bob do that part for you--well, it makes me cranky. In fact this touched enough of a nerve that I'll make an entire separate post about it. You can read that now or later-- TL:DR, having Bob doing all the start-up work for your assignment is not okay.
Bob-Assisted Editing. In this option, Bob reads over your work and tells you what to fix. He can't add whole new sections, but he can do anything else to "improve the quality" of your work.
Bob for Specified Task Completion. Maybe when I gave you the assignment I said, "Go get Bob to make your charts" or "Have Bob collect all your research materials" or some other specified task. Why this is Level 3 when it seems like potentially the least objectionable use of Bob I do not know. But this is probably a good time to mention that while Bob is smart, he also has a serious drinking problem, and whatever task he completes for you, you'd better check over carefully, because I'm going to hold you responsible for the part of the assignment that I told you to have Bob do.
Full Bob Use with Student Oversight. In this option, you just have Bob do the assignment for you. How having Bob as your "co-pilot" as a way to enhance your creativity is beyond me; maybe the creativity part comes when you explain why you should get credit for Bob's work. If Bob screws anything up, it's on you, though I cannot for the life of me figure out what I am assessing when I give you a grade for Bob's work.
In fact, that's a problem for most of this. I am trying to assess certain skills and/or knowledge of you, the student. Bob isn't even in my school, let alone in my gradebook. So how do I award a grade to student based on Bob's work?
If you agree that the thought of a student running off to have neighbor Bob complete some-to-all of that student's assignment seems like an ethical and assessment problem, then someone explain to me why using AI is any different or better. I have no doubt that it will be some-to-all degree of difficulty to keep some students from getting Bob to help them complete their assignment, but that doesn't mean I should create a formal structure for how much of what kind of cheating they will be using in my class.
*I generally default to "Pat" or "Sam" or other gender non-specific names, but "Bob" is objectively more funny.
No comments:
Post a Comment