MENU

AI-Proofing Assignments

August 22, 2024

Sorry, but I overcommitted myself again this past spring 2024.  I will do better sending out MTI Teaching tips this fall 2024 semester.  No promises for spring 2025.

Some Public Service Announcements

From Katy Little at TILT – We invite you to enroll in some of the most popular programs we offer, our Best Practices in Teaching (BPiT) courses!

These three-week courses are designed to enrich your teaching skills while operating asynchronously in Canvas, with most of the work happening on your own schedule. During these courses, you’ll have the opportunity to explore, apply, and discuss research-based teaching approaches based on the Teaching Effectiveness Framework. Plan to dedicate 4-5 hours per week to assignments, discussions, readings, and meaningful interactions with fellow educators across campus.

We MTI Coordinators would first like to remind you of the First Four Weeks Initiative (FFW) and share with you resources and tips you can use at the beginning of the semester. You can find a brief overview and a few best practices from the First Four Weeks Initiative that will help your students, whether in-person or online, succeed in your courses.

There are many possibilities for this MTI Tip.  For this week I decided on generative AI (ChatGPT, etc.). This tip is based on information posted by Joseph Brown, Director of the Academic Integrity Office, CSU TILT.  My department is wrestling with how to approach the use of generative AI in our senior thesis course.  While looking for information and guidance I ran across Dr. Brown’s very useful blog on the TILT website.  Rather than dump all the information on you at once I will go through each of his five posts in the next several weeks.

How to AI-Proof Assignments by Joseph Brown

I get a lot of questions about how to make assignments resistant to AI cheating. I understand where this question comes from. Our faculty are creative, hard-working, intelligent professionals and they believe that challenges can be worked through. I believe that too, but I’m also worried that we keep thinking that successfully managing AI is a short-term challenge as we wait for a technical solution or a practice that will eliminate the possibility of cheating. The AI thing is the definition of a moving target. Approaches from a year ago aren’t as relevant now. Instead, we are in the so-called “messy middle” of a profound shift in how people access, use, and produce information. 

First, there are no “AI-proof” assignments. There are assignment/ assessment design approaches that, seemingly, make AI use seem irrelevant and/or silly. They aren’t perfect. And they are continually evolving. For example, it might seem odd to us that a student would go to AI to produce a response to a reflective question like “What discussion board post or in-class comment by a peer did you find most thought-provoking this week?” Then again, the psychology of students in this moment probably eludes most of us. I don’t think we should be that surprised if they do ask AI for such a response.

However, I do want to share what I’m learning in the various AI and Academic Integrity/ Teaching & Learning workshops I’ve attended. I share these not because I think “best practices” are a magic wand that we can wave and resolve these challenges, but because I think sharing the ideas generates creativity and innovation. Here are some examples from notes I took while attending a workshop led by Notre Dame’s James Lang last semester:

Reflective Assignments

Self-Assessments:

  • Assignment wrappers- Write an analysis of your performance on the last exam/ assignment. What did you do well? What didn’t you do well? How will you improve?
  • Learning Journeys (longer assessment): Use five of your low-stakes assessments this semester (quizzes, discussion board posts, chat comments, group work assignments, etc.) to describe your learning journey this semester. 
  • Questions for Discussion Boards:
    • Which discussion board post or in-class comment by a peer did you find most thought-provoking? What did you learn from it?
    • Review the syllabus and your course notes. Which discussion or class period was most helpful to you this semester? How did it change your thinking?
    • Review the course lectures or readings. Which one changed your understanding of the course material the most?

Alternative Exam Questions (according to mathematician Franci Su):

  • What mathematical ideas are you curious to know more about as a result of taking this class?
  • Take one homework problem you have worked on this semester that you struggled to understand and solve and explain how the struggle itself was valuable.
  • How has your mathematical imagination been enhanced as a result of taking this class? Give at least three examples. 

Enhanced Assignments

Next Generation Genres: What are the next generation assignment genres that will inspire student learning and help maintain academic integrity in student learning? (credit: Jessica Singer Early). 

  • Work + Creator’s Statement
    • Creative work and an Analytical Statement
    • Traditional Essay and a Writer’s Statement
    • Presentations and a Speaker’s Statement
    • “What did you find energizing in creating this work?”; “Where did you struggle? How did you respond to the challenge?”; “How well did you achieve your vision?”
  • AI- Developed Work+ Learner’s Statement
    • How did you use AI in the completion of this assignment?
    • Present and explain the initial prompts you used, with an explanation of your reasoning behind any further prompts.
    • What did you learn from your experience working with AI?

Again, these examples come from my notes on that workshop and, as you can see, they represent an attempt to reformulate traditional assignments so that the focus is on learning, not necessarily preventing cheating. Lang especially emphasized this last point and I think it’s worth keeping in mind as we all approach our courses this term.

This isn’t an exhaustive list, and it may be irrelevant in a short time. However, I hope these examples give you some insight on how professionals in higher ed are responding to a generational challenge and consistently moving target.

Food for thought.  I have noted among my colleagues a broad range in opinion about the criticality of preventing cheating.  I have noted a similar diversity regarding the ability to detect generative AI use and the need to prevent that use.  I have no concerns about generative AI use in any of the courses I teach other than Senior Thesis.  So far, I have not run across any indication that students have used generative AI while writing their theses.  If any students have it was of no help at all.  I welcome your feedback and suggestions on this or any other teaching topic. I am looking forward to sharing another of Joseph Brown’s blogs next week.

Cheers, Paul