When they say “AI helped me,” skip the trap question

When your child says, “AI helped me,” a trap question is waiting in your mouth: Did it do the work for you?

That question is fair. It is also a courtroom. Many children hear it as “confess.” The useful facts — what they asked, what they changed, what they still do not understand — stay behind the defence.

Try three calmer questions instead.

Why the trap question backfires

“Did it do the work for you?” has one acceptable answer: no. Everything else sounds like trouble. A child who used AI to unpack a wordy paragraph and then wrote their own sentences has no clean place to stand. A child who pasted a whole answer will just say “no” anyway.

You still care about honesty. You still care whether the assignment is theirs. The shift is how you find out. Common Sense Media’s chatbot guidance is built on this idea: you do not need to understand every assignment to have a useful conversation about how they used the tool. Questions like “What did you figure out yourself before you asked AI?” and “Where did you disagree with what the AI said?” do more work than a yes-or-no (AI chatbots and your child).

The goal is a picture of the process, not a verdict in the doorway.

The three questions

What did you ask it? This is the prompt. A vague “help with science” is different from “explain photosynthesis in words a ten-year-old can say back.” You learn whether they used the tool as a search box, a tutor, or a ghostwriter.

What did you keep, change or reject? This is the edit. If they cannot name a single change, that is information. If they can point to a sentence they rewrote or a step they ignored, that is also information — and usually better news.

Which part feels most like yours? This is ownership. It is harder to fake. A child who did the thinking can usually tap the page. A child who did not will stall. You do not need to pounce. The stall is your cue to sit down and rebuild the work together.

Those three questions move the conversation away from catching them out and toward how they made decisions. You still get a clearer picture of how AI was used. They get practice explaining their choices — which is the skill a teacher can mark, and a parent can hear.

How this sounds on a school night

Keep it short. Bags still need emptying.

• “Show me the question you typed.”
• “What did you throw away?”
• “Read me the bit that is most you.”

If they used AI as a coach — hints, a worked example, a definition — you can say so out loud: “That’s a tool. Now show me you can do the next one without it.” Common Sense’s review of AI in schools makes the same split: tools that ask questions and give hints support learning better than tools that hand over answers (AI chatbots and your child).

If the page is mostly the chatbot’s voice, do not turn it into a trial. Close the laptop. Ask them to explain the idea on scrap paper. Start again from the bit they do understand. The consequence is redrafting, not a speech about the future of honesty.

Check the school’s AI rules once, then use the school’s words at home. “Our teacher says you can use it to plan, not to write the final paragraph” is clearer than a house lecture about plagiarism.

What you are actually teaching

You are teaching that AI use is something they can talk about, not something they hide. The child who can narrate their choices will still need your eye. They will not need a lie ready at the door.

Use the same three questions for a Pip’s Lab build, a story, or a poster. The tool changes. The habit does not: ask, edit, own.

When they say “AI helped me,” try to hear a report, not a confession. Then ask the three questions — and listen to the answers longer than you talk.