A Professor Set a Genius AI Trap for His Students. 32 Out of 35 Fell for It.

AI was supposed to make teaching easier, but instead teachers are spending more time policing student work, detecting AI-generated essays, and figuring out how to keep students learning. Remember when everyone said artificial intelligence was going to make teachers’ lives easier? And to be fair, it does….. somewhat. AI can crank out lesson plans, generate activities, differentiate assignments, and provide instant feedback in a matter of seconds.

But there’s one little problem: How do we get students to actually learn when AI can do so much of the work for them?

Suddenly, we aren’t just teaching and grading. We’re becoming AI inspectors, trying to determine whether the essay sitting in front of us was actually written by our student or a chatbot. We’re comparing writing samples, questioning vocabulary, and wondering how the kid who normally writes, “The Industrial Revolution was good,” suddenly sounds Harvard-educated.

Well, one professor had enough and came up with a great idea to prove AI was being used for more than just ideas.

The Professor Who Set a Trap for AI

A history professor at Alcorn State University recently went viral after revealing a rather clever way he caught students using AI on a midterm.

He inserted hidden text into the assignment using white font on a white background. Students couldn’t see it when they read the assignment normally, but when they copied and pasted the entire prompt into an AI chatbot, the hidden instruction went along with it.

The instruction told the AI to insert a completely random word into the response: “Madagascar.”

The assignment was about the Industrial Revolution. According to the professor, 32 of the 35 students submitted responses containing the bizarre hidden instruction. He said the students apparently didn’t even proofread the AI-generated responses before turning them in. That’s honestly not surprising.

Now We’re Analyzing Essays for AI

One of the biggest changes AI has brought to teaching is that teachers can no longer simply read an essay and focus on the student’s ideas. We’re now reading it with a completely different question in the back of our minds: Did the student actually write this?

We’re looking for writing that sounds “too good” for our students’ abilities. This becomes very difficult because it is so subjective. The frustrating part is that we want to trust our students instead of accusing someone of cheating.

We should be spending our time giving students feedback on their ideas, helping them become better writers, and encouraging them to develop their own voice. Instead, we’re analyzing word choice and judging writing ability. Teachers aren’t just grading essays anymore. We’re investigating them.

And Then We Were Told to Use AI Detectors

The solution to this problem, we were told, was supposed to be AI detectors.

Just run the essay through a program and let it tell us whether the student used AI. Sounds easy, right? Except it isn’t, and it isn’t accurate. AI detectors aren’t reliable enough to give teachers a clear yes-or-no answer. They can flag writing that was actually written by a student, while AI-generated writing can sometimes slip right through. A teacher can end up accusing a student of cheating based on a percentage that isn’t proof of anything.

So now we’re stuck in an impossible position. We’re expected to catch students who use AI, but the tools designed to help us catch them aren’t always accurate. We can’t simply look at a detector score and say, “Yep, got em.” That means we’re back to analyzing the essay ourselves, comparing it to previous work, questioning the student, and coming up with schemes like the “Madagascar Professor.”

So much for saving teachers time.

AI gave us a tool to save time, then created a problem that takes even more of it away from us. And somehow, the teacher is still the one responsible for figuring it all out. Of course we are. We always are.

So What Is the Solution?

Where does this leave teachers? One possible solution is taking more of the writing process back into the classroom. Instead of sending students home with an essay and hoping they actually wrote it, teachers can have students brainstorm, write, and revise during class. That makes it much easier to see their thinking and their writing process firsthand. Limit Chromebooks. Bring paper-and-pencil work back as the norm. Of course, this takes lots and lots of classroom time, but it ensures students are actually doing the work.

Maybe the answer isn’t to ban AI or spend our days trying to catch students using it. Maybe we need to teach students how to use AI responsibly while still making sure they are doing the thinking themselves. That’s the tough part. Human behavior wants to save time, but we don’t want to eliminate thinking altogether. AI can be an incredible tool for teachers and students, but it shouldn’t replace learning.

Because at the end of the day, that’s what this is really about. We’re not worried that students are using AI. We’re worried that they’re using it instead of learning.