I Had a Cheating Problem, But It Wasn’t What I Thought: AI in the Classroom

Getting my students to do homework this past school year was tough.

They didn’t see the value in it, and honestly, why would they? They could plug the question into AI, get the answer in seconds, and move on. By the end of the year, it got so bad in one class that some days I stopped really checking homework altogether and just gave a completion grade. They all had answers. Most of them were even correct.

The problem was that they didn’t understand any of it.

We would spend class time going over homework, but there was barely a point. Students already had the answers in front of them, but many of them could not explain the thinking behind those answers or how they got there in the first place.

By May, I knew I needed a new plan for next year.

Not just a stricter policy. Not better AI detectors. A completely different approach to assignments and accountability. I needed a system that actually required students to think and one that placed value on doing the work, not just turning something in.

But that raises a bigger question:

How do you get students to see value in the process when technology can instantly give them the answer?

How AI Changed Homework in My Math Classroom

This year, classwork and homework assignments were graded mostly on completion.

Technically, I was supposed to check for shown work, but on many assignments, there really was not much work to show. That eventually led me to mostly check whether students had answers written down.

For some groups of students, maybe that system works. But for the majority of my Geometry students, it meant they were not actually engaging with the math.

And AI made that problem even more obvious.

Students who knew they were supposed to “show work” often came in with pages full of steps, but it was obvious the work was AI-generated. It did not match the methods we had learned in class. Sometimes it skipped reasoning entirely. Other times, it assumed information that students themselves could not explain.

Sure, technically, there was work on the page.

But most of the time, it was not evidence of understanding. It was evidence that they had checked the answer.

That was the moment I realized my cheating problem was not really about cheating.

It was about a system that no longer measured learning.

Why Traditional Homework Assignments Are Failing in the AI Era

I would love for my students to naturally understand the value of struggling through a problem and learning from mistakes.

But I can also be realistic.

When students have access to instant answers, it is hard to compete with convenience. Instant gratification will almost always win over long-term growth, especially for teenagers.

In an ideal world, every student walks into class motivated to learn and determined to be successful. In reality, many students arrive having already made up their minds about school, and especially about math.

They tell themselves:

• “I’m just bad at math.”

• “I’ll never use this.”

• “It doesn’t matter.”

• “AI can do it anyway.”

One teacher is probably not going to completely undo years of frustration or negative experiences with school.

But we can make small shifts.

We can create classrooms where thinking matters more than speed, where students are expected to explain ideas instead of just produce answers, and where completing the work actually has meaning.

How I'm Redesigning Math Assignments for an AI Era Classroom

The biggest thing I realized this year is that assignments that worked five years ago simply do not work the same way in an AI era.

Many traditional assignments are too easy to outsource and do not actually check for understanding.

So this summer, I am reworking both my instruction and my assignments.

For example, instead of giving students 10 equations to solve, an Algebra assignment might look something like this:

• 2 problems where students identify and fix an error

• 2 problems where students explain the process before solving

• 1 application or word problem

• 5 traditional equations to solve

The math itself is not necessarily harder.

But the thinking is deeper.

Students have to explain, analyze, justify, and apply what they know instead of simply finding an answer online and copying it down.

Will this completely solve the problem? Probably not.

Students will still try shortcuts. AI is not going away. But I think the goal now is not to eliminate AI. The goal is to create assignments where understanding is harder to fake.

Creating a Classroom Culture Where Students Actually Think

The other piece of this is culture.

Students need to see meaning behind doing the work, and part of that starts with accountability.

Next year, I want students to take more ownership over checking their answers, explaining their thinking, and defending their reasoning. Some of the classes where I learned the most as a student were the ones where I had to be prepared to explain my homework out loud.

So I plan on bringing some of that into my classroom.

Not every day. Not for every question.

But if students know they might be asked to explain an answer, model a problem, or defend their reasoning in class, it changes the way they approach the work.

At least, that is the hope.

Because the more I think about it, the issue was never really that students suddenly became cheaters.

The issue is that education changed, technology changed, and many of our assignments did not change with it.

Over the summer, I’ll be creating and testing more AI-era math assignments focused on reasoning, explanation, and student thinking. As I build them, I’ll be sharing them in my Teachers Pay Teachers store for teachers who are trying to navigate the same challenges in their own classrooms.

You can find those resources here: Bugged About Math on TPT

And if you want more context on how this school year shaped a lot of these reflections, you can also read my end-of-year reflection.