On “AI-Proofing” Your Assignments

I need to get something off my chest. I’ve been struggling and I know that many of my colleagues have been struggling. Sharing those struggles online has been its own struggle. I’m talking about students submitting work that is generated by large language models (LLM) and my efforts to deal with “AI-proofing” assignments.

When I share my frustrations online, or commiserate with other educators who are also frustrated, I am often met with suggestions that the work I assign my learners is somehow antiquated, traditional, not meaningful, needs to include them more, or whatever their hot take on the issue might be. Social media (particularly Twitter) is likely not the place to have these conversations, but it’s what I have.

Simply sharing my struggles is often met with an assumption that it must somehow be because the assignments I’m giving my learners just aren’t relevant; or they are not soliciting their own insights, opinions, culture, passions…the list goes on. It’s usually from a person from the world of higher ed who somehow has a handle on the challenges that K12 teachers face, and if we would just do things differently (better) then we wouldn’t have such a problem with AI submissions because students would see value in the authentic learning experiences!

I find it terribly offensive. I take great pride in my work. I also take it very seriously. As a teacher who writes most of my curriculum, I have always made great efforts to find ways to help my students relate to the instructional content in ways that are personal and meaningful. My course surveys are filled with commentary from so many students about their experience and the value they placed on my course. It does the things I wanted it to do…in many cases.

I find it hilarious that an academic researcher, or someone with no experience teaching learning populations that struggle, would somehow find it just as easy as finding more ways to include them in their work.

Here’s the thing, it doesn’t matter.

There is a big difference between someone (an adult) who is paying for a course, working to earn a college degree, and someone who is in high school, who has all kinds of good reasons for feeling disengaged, or has hated school for many years. I can tell you, as someone who grades multiple AI submissions a day, that it does not matter what you do to your assignments in some cases. Some students will still try to use AI to do the work for them. I get it. I probably would have done it as well. They will often use LLM to compose responses to questions that are meant to evoke some personal connection or insight into the material. They have it make that up for them too! Some will take even the shortest and easiest assignment which is just a simple check for their opinion and reflection and use AI for that as well. It’s baffling to me because it would take longer than just writing a couple sentences, but here we are; it’s happening.

I sympathize with teachers who may be using a vendor curriculum or teach in a setting where they have to teach the same things as the person across the hall from them and they are unable to deviate from the district-approved plan. That’s rough. They’re really in a no-win situation.

Is it futile to try to find ways to AI-proof your assignments? I don’t know yet, but it doesn’t mean I’m going to stop trying. I just don’t need the suggestions that what I’m asking my learners to do doesn’t matter.

By Jon Fila

Jon is a Teacher, Author, Speaker, Consultant, AI Strategist who focuses on equity issues (accessibility, racial, gender), accessibility and Open Educational Resources. He has worked in education for over twenty-five years. He has served as an Innovation Coach, Curriculum Coordinator, PD Specialist, Department Chair, and has worked on aspects of Strategic Planning and has facilitated those groups on topics of Equity. He has been making his own stuff for a long time and shares whatever he can.

Leave a comment

Your email address will not be published. Required fields are marked *