The "Explain It to Me" Test: Did Your Kid Actually Learn It?
If your child used AI for homework, one simple question tells you everything: can they explain their work in their own words? Here's how to use it.
Marcus sat down at the kitchen table to review his eleven-year-old's book report. It was polished. Coherent paragraphs, a clear thesis, specific examples from the text. He was genuinely impressed. Then he asked a simple follow-up question: "So what happens to the main character at the end?"
His son looked at the paper.
"I mean, without looking at it," Marcus said. "Just tell me."
A long pause. A shrug. "I don't really remember."
The report was excellent. His son had learned almost nothing from writing it.
This gap slips past rubrics, evades teachers with thirty students to grade, and defeats AI detection tools entirely. The output looks like learning happened. The child knows that it didn't. One simple practice, applied consistently at home, closes the gap faster than any policy or punishment: ask them to explain it.
Why Explanation Is the Real Test
When a student genuinely works through a topic, they build a mental model of it. Not just words on a page, but an internal structure of understanding that lets them answer questions they haven't rehearsed, make connections to things they already know, and talk about the material in their own voice.
When a student delegates thinking to AI and submits the output, they have the product without the process. The mental model never forms. They can describe what the paper says because it is right in front of them. They cannot explain the material because they never understood it.
This is not a new problem. Students have always found ways to produce work without learning. But AI has made the gap between output quality and actual understanding wider than at any previous point. A 2025 study by Michael Gerlich, published in MDPI's Societies journal, found a strong negative correlation of 0.75 between frequent AI tool reliance and students' critical thinking scores, mediated by cognitive offloading, the tendency to outsource mental effort to a tool. That is a correlation, not proof of causation, and a correction to the study was published in September 2025; the core findings remain intact. The students most likely to offload were also the least able to demonstrate understanding when the tool was removed from the equation.
The "explain it to me" test removes the tool from the equation. Instantly.
What the Test Actually Involves
The test is a simple conversation. It requires only that the child talk about their work as if the paper does not exist, without reciting facts or performing under pressure.
The format adapts depending on what your child just worked on, but the core move is the same. Close the laptop or turn over the paper and ask them to describe the piece from memory: what it is about, what the main point is, and one thing they found genuinely interesting. Then ask a follow-up that the paper answers but that requires understanding to get right. For research work, ask where the information came from and how they know it is accurate, which asks them to trace a claim back to a source out loud. For creative or problem-solving work, ask how they decided to approach it and what choices they made along the way. Five minutes of conversation is enough to know whether understanding is there.
What You Are Listening For
You are not grading. You are not looking for perfect articulation or complete recall. You are listening for evidence that a mental model exists.
A child who understood their work will stumble a little, use approximate language, get some details slightly wrong, and still clearly know what they are talking about. They will say things like "I think it was because..." and "the part I found most interesting was..." They will be able to answer a follow-up question that is not in the paper, because understanding generalizes.
A child who did not engage with the work will produce a blank expression, reach for the paper, or give answers that are clearly word-for-word from the text they just submitted. They are retrieving, not understanding. The difference is audible within about sixty seconds.
Making It a Habit, Not a Trap
The instinct, when you discover a child did not actually engage with their work, is to treat it as a disciplinary event. That instinct is understandable, but it tends to make the next conversation harder. If children associate "explain it to me" with being caught and punished, they will approach it defensively rather than honestly.
The practice works better as a genuine curiosity habit, applied consistently and to work you already know was done well. Ask your eight-year-old to tell you about the picture they drew before you can see that a sibling helped. Ask your teenager about an essay you know they wrote entirely on their own. When the conversation is low-stakes and regular, it becomes normal. Then, when the work was actually produced with AI and understanding is absent, the gap reveals itself in the same neutral, non-accusatory context.
The goal is to close the gap between what was submitted and what was learned.
Why Schools Cannot Do This for You
A teacher with twenty-eight students grading sixty papers cannot have a five-minute oral conversation about every submission. The structural reality of classroom instruction means that output quality is what gets graded, because it is what is visible and practical to assess.
The "explain it to me" test is something only a parent or small-group educator can do consistently. It requires proximity, time, and a relationship where the child does not feel they are being examined. It is one of the genuine advantages supplemental and homeschool families have over traditional classroom instruction, not because schools are failing, but because one-on-one conversation is simply not scalable at a classroom level.
The families who build this habit early, before AI-produced work becomes habitual, are building something more durable than any homework policy: a child who knows that producing work and understanding work are not the same thing, and that the second one is what actually matters.
For a deeper look at why AI can produce polished outputs without student understanding, see our post on cognitive offloading and how AI dependency forms.
Sources cited in this post:
- Gerlich, M. (2025). "AI Tools in Society: Impacts on Cognitive Offloading and the Future of Critical Thinking." Societies. MDPI. https://doi.org/10.3390/soc15010006