What did we do when students stopped thinking?

While reading assignments last term, I noticed something odd. The arguments were sound, the language was fluent, the sources were all in place. But when I asked the same students in class "why did you think that?", they could not answer.

The problem was not that the assignments were bad. The problem was that the assignments were good and the student no longer had anything to do with them. They had used AI instead of thinking.

Why banning does not work

The first reaction is usually a ban: "using AI in assignments is not allowed." This has two problems. First, it cannot be enforced; detection tools are not reliable, and a false accusation costs more than a missed case of cheating. Second, and more important, it deprives students of a tool they will use every day after they graduate.

The real issue is not the tool itself but the point in the student's thinking process at which it comes in.

Rebuilding the flow of the course

I tried three steps; the three work together.

Your own sentence first. Before opening the AI, the student writes down their own idea in two or three sentences. This small step changed everything. A student who has a sentence of their own can compare the text the AI produces with their own idea; a student who does not simply accepts the text as it comes.

Use AI for the counterargument. Not to have it write the assignment, but to have it produce objections to your own view. This one-sentence instruction turns the tool from an author into a debating partner. The student has to either refute the objection or change their mind; both are the work of thinking.

Take ethics into class discussion. The question "what does it mean to own an idea?" is not only a matter for the philosophy class. In a class that discusses this openly, you do not have to impose the rule from outside.

Rethinking what we measure

These three steps worked, but the real change was in assessment. Instead of grading the submitted text, I began asking about the decisions the student made along the way: at what point did you consult the AI, what did it suggest, why did you accept or reject it?

If AI thinks in your place, you are not learning, you are only handing something in. The student we want to educate is not the one who uses AI uncritically, but the one who knows when to trust it and when to object.