When a Student Says, "I Used ChatGPT to Revise"

Mar 2026

Yesterday, I was giving feedback on a high school student’s written CER response when I asked a question that is becoming more common in academic conversations: “Did you use AI anywhere in this process?” He answered easily and honestly. Yes, he said. He had used ChatGPT for revision.

That answer is more revealing than it first appears.

“Revision” can mean many things. It can mean catching repetition, tightening a sentence, or noticing that a paragraph wanders. It can also mean asking a tool to sharpen a claim, reorganize an argument, suggest evidence, or supply language that the student could not have produced independently. Those are very different kinds of help, but they are increasingly being grouped together under one harmless-sounding word.

The larger context matters. Student use of generative AI for schoolwork is no longer occasional or marginal. Pew Research found that 54% of U.S. teens ages 13 to 17 say they have used AI chatbots for help with schoolwork, and 35% say they have used them to edit something they wrote. RAND reported that AI use for homework among middle school, high school, and college students rose from 48% to 62% between May and December 2025. College Board likewise found that 84% of high school students reported using generative AI for schoolwork in May 2025, with half using it to brainstorm, revise essays, or conduct research. RAND also found that 67% of students said using AI for schoolwork harms critical thinking.

Those numbers help explain why this conversation can no longer be reduced to simple warnings about cheating. Many students are not turning to AI because they are lazy. They are turning to it because they are overwhelmed, uncertain, pressed for time, or afraid of getting the work wrong. Pew found that 68% of teens say they feel a great deal or a fair amount of pressure to get good grades. In that environment, a tool that offers immediate structure and fluent language can feel less like an ethical decision and more like relief.

That is precisely why the line matters so much.

In a CER response, the deepest academic work is not the final polish. It is the thinking that comes before polish: forming a defensible claim, selecting relevant evidence, and explaining how that evidence supports the claim. If a student drafts those moves independently and then uses a tool to notice awkward wording or repeated phrasing, that is one kind of assistance. If the tool is quietly helping to generate the reasoning itself, we are in a different territory altogether.

This is where I think many adults need a more precise question. The most useful question is not simply, “Did the student use AI?” It is, “What part of the thinking remained the student’s?”

That question changes everything.

A finished paragraph can now look stronger than the understanding behind it. A response can appear organized, clear, and even insightful while concealing how much of the intellectual work was externally supplied. The product, in other words, is no longer enough. We have to look more closely at process, authorship, and transfer.

When I want to understand whether AI has supported learning or quietly stood in for it, I look for a few things. Can the student explain, in conversation, why this piece of evidence supports this claim? Can the student respond to a related prompt without relying on borrowed structure? Can the student tell me which revisions clarified existing thinking and which ones introduced ideas that were not there before? Those moments usually tell the truth more quickly than the polished submission does.

None of this means AI has no place in student work. It does. Students can use it to clarify directions, compare models, identify mechanical errors, or generate questions that help them rethink a draft they have already written. But those uses require boundaries. If revision becomes a vague catch-all, students can slip from support into substitution without fully realizing it themselves.

That is why I believe schools and families need a more disciplined conversation about AI and revision. Students need clearer language for describing how they used a tool. Teachers need assignments designed so that the important thinking is still visible somewhere along the way. Parents need a framework that goes beyond “AI is good” or “AI is bad” and instead asks whether the tool is extending a student’s reasoning or replacing it.

The question that matters most is not whether a tool touched the work. The question is whether the student still had to do the part of the work that builds understanding.

That is the boundary I keep returning to.

If AI helps a student see their own thinking more clearly, it may have genuine value. If it begins doing the reasoning the student most needs to practice, the work may improve on the surface while the learning underneath becomes thinner.

And that is a trade most students cannot afford to make.

---

Sources

Pew Research Center. How Teens Use and View AI. February 24, 2026.

RAND. More Students Use AI for Homework and Believe It Harms Critical Thinking. March 17, 2026.

College Board. New Research: Majority of High School Students Use Generative AI for Schoolwork. October 6, 2025.

Pew Research Center. Pressures Teens Are Facing. March 13, 2025.