"Stoplight" approach to AI use in writing assignments

Like you, my Inbox is bombarded with messages about AI. I try to keep up, to stay especially focused on the writing instruction-centric ones, but it seems that only an AI-based triage tool can help me with the deluge...

Teachers do keep productively discussing the challenges of student use of AI, especially in writing instruction. This term, I am team-teaching an Honors seminar Contrasts in AI: Now & Future, part of Drexel's overall Honors College theme this year: Black & White. The class has been a great experience, particularly working with my colleague, a computer and electrical engineering professor.

Our course is about AI, and in addition to many informal assignments such as reading notes and quizzes, we also have three short papers/projects. We talked about this a lot in the summer, and we eventually developed a "stoplight" approach that we thought not only fit well with our course subject matter but also helped us address the challenges of requiring short papers in this current AI educational environment. Lightly edited from the syllabus, here is what we asked of students:

You will write three short, 500- to 1,000-word papers/projects in the course, due Weeks 4, 8, and 11. We’ll discuss each in class when it is assigned, but in general you’ll follow these guidelines:

1. For the theme/thesis of each paper/project, you will develop a contrasting perspective/view of some aspect(s) of AI, building on the course materials and conversations.
2. Each paper/project will be different in its use of AI, based on a “stoplight approach”:
• Project one is RED: You cannot use any AI, including grammar checkers like Grammarly. It may be impossible to avoid some built-in tools in Word and Google Docs.
• Project two is YELLOW: You can use some components of AI to help you develop your ideas and deliver them. You might think of writing on a kind of scale of complexity or a continuum, starting with the word, working through the sentence, then paragraph, then overall concept/purpose. (We missed a few steps in between for sure.) For “Yellow,” please do use AI bots/apps in your project, but only do so up to the paragraph “level.” We ask that you still not use AI for invention, conceptual, or organizational matters, but you should feel free to “partner” with ChatGPT, Copilot, etc. to help you at the sentence- or word-level, perhaps in particular with grammar and mechanics.
• Project three is GREEN: You can freely use AI as a partner to help you write your paper. Note that you cannot just submit the raw text generated by an AI app/tool, but you can certainly go through an evolution of prompts in presenting your final paper.
3. For each paper, you will include at the end a one-page metacommentary about how the use of or the constraints on the use of AI affected your development of your ideas. There are alternative ways you could present this metacommentary that we’ll discuss in class. Note this is different from the AI statement described in the Course Policies below.

So far, we've received the red and yellow papers/projects. This class of bright students have written strong, thought-provoking projects, but, boy, has the hidden gold been in the metacommentaries. 

In these end reflections, students have commented in considerable depth about the constraints of the red and yellow categories, how they navigated what those categories have meant to their writing process, and in some cases how they hadn't realized how much they had become reliant on AI tools in their writing. (Note the metacommentaries are part of the grade, a deliberate move we made so as to reward their thoughtful work.)

I want to reiterate that our class is about AI: Issues surrounding AI are part of the course subject matter. That certainly helped lead us to developing a "medium is the message"-type approach to the assignments.

However, including these detailed metacommentary components has helped us with our writing pedagogy. While plagiarism and authenticity weren't primary targets, the metacommentaries have encouraged authenticity while also providing them with a platform to discuss their own writing and thinking processes, in line with decades of composition work about the value of such writing.

As instructors, we are eager to receive paper/project three next week, not only to see their final thoughts about AI contrasts but to read how the open green category differed for them in their use of AI in their writing.