Seven in Ten Teens Use AI for Their Homework. Four in Ten Admit They Are Missing Real Learning. Here Is the One Skill Worth Teaching Before Your Kid Stops Knowing the Difference.

My oldest came home with an A on her biology worksheet last week. I knew she had used AI to help with it. I asked her to explain one of her answers to me. She stared at me. She genuinely could not. She had the grade but not the knowledge. That moment felt like a preview of a gap I did not have a name for yet. Then a new survey from Common Sense Media dropped, and I found the name. Seven in ten teens use AI for schoolwork.^[1] Four in ten say they are missing real learning because of it.^[1]

Seven in ten American teenagers now use AI for their schoolwork, and a significant share of them are using it in ways that get them answers without ensuring they got the understanding too.

Common Sense Media published a nationally representative survey of more than 1,000 teenagers on August 18, 2026, called “Teens in the AI Era: Schoolwork and Skills That Matter.”^[1] The headline number is 70 percent using AI for schoolwork. But the more interesting number is underneath it: 63 percent of those teens say they use AI to get answers, whether they paste the response in directly, rewrite it to sound like themselves, or blend it with what they already know.^[1]

That last group, the 35 percent who improve AI answers with their own knowledge, sounds the healthiest. But improving an AI answer is not the same as generating your own. You are starting from a floor someone else built. A kid who edits an AI response has done less cognitive work than a kid who wrote a rough draft and then revised it. Those are meaningfully different activities. One of them transfers to a closed-book exam. The other does not.

A study of 26,811 secondary school students tracked over 30 months illustrated exactly that gap.^[3] When students used AI for homework, their assignment scores rose 18 percent and they finished assignments about 19 minutes faster. On closed-book exams taken six months later, the same students scored 20 percent lower than they had before AI adoption.^[3] The assignments got better. The understanding did not. More students getting better grades on worksheets, fewer who can explain what they learned. That is the trajectory the survey found in every school that adopted AI tools without teaching students how to use them well.

The issue is not that your kid is using AI. The issue is that most of them cannot tell the difference between getting an answer and learning one, and almost no one is helping them develop that distinction.

Two-thirds of the teens surveyed say they believe AI helps them understand material, not just complete it faster.^[1] That belief is widespread, and it is probably mostly sincere. But when the same survey asked whether using AI made them miss learning opportunities, 40 percent said yes.^[1] Both of those things cannot be entirely true at the same time. Either AI is helping students actually learn, or it is helping them skip the parts of learning that feel uncomfortable. Most teens are not sure which is happening to them. And most of them do not have the tools to figure it out.

Only 30 percent of teens say a teacher has ever talked to them about how to use AI safely or appropriately.^[1] That means the majority of students are making these decisions on their own, without guidance, in real time, often while trying to finish homework after a long day. The choice they make in that moment, whether to engage with the material or let AI handle it, is shaping habits that will follow them into settings where closed-book and no-AI tests are the measure of what they know.

Teachers see the gap widening. In a nationally representative poll of 545 K-12 teachers by NPR and Ipsos, 57 percent said AI has made it harder for them to assess what students actually know.^[2] They watch students turn in polished work and then struggle to explain it during class. That gap between what the assignment shows and what the student understands is not new. AI has made it wider and faster to create.

The skill is Learning Honesty, and it comes down to one question your child should be able to answer after any AI-assisted task: Can I explain this without AI?

Learning Honesty is not about banning AI or treating every AI interaction as cheating. My oldest is going to use AI for the rest of her life. What I want her to be able to do is accurately judge whether she learned something or only produced something. Those two things feel identical in the moment. They feel very different on a test three weeks later, or in a college interview, or in a job where no one is doing the thinking for her.

The 30-month study put numbers on that feeling.^[3] When AI was available, performance looked great. When it was not, the understanding gap became visible in a closed-book room where none of the AI-assisted preparation transferred. The assignments improved. The student did not. Learning Honesty is the habit that closes that gap. It means checking, honestly, whether you could reproduce the thinking without the tool right now. And if the answer is no, not accepting that as a finished homework session.

This is different from the critical thinking skill I have covered before, which is about evaluating sources and information quality. Learning Honesty is about evaluating your own understanding. Your kid can read a perfectly accurate, well-sourced AI response, internalize nothing meaningful from it, and still complete the assignment. Information quality is not the same as comprehension. This skill is internal. It is about knowing what you actually know.

A child building Learning Honesty will volunteer to explain what they learned without being asked, and will tell you honestly when they used AI and still do not fully understand the material.

One clear sign is that your kid re-explains things in their own words without prompting. My middle daughter does this naturally. She talks through what she learned at dinner, in her own clumsy 9-year-old version of it, and I can hear whether she actually got it. That is Learning Honesty in practice. She is checking her own understanding out loud. She does not always get it right. But she knows the difference between repeating words she read and explaining something she grasped, and that distinction is what matters.

Another sign is that your kid pushes back on AI answers sometimes. They say things like “that does not seem right” or “I thought it was different.” That happens when they have enough independent knowledge to compare the AI response against. A kid who does all of their learning through AI has no baseline to push back from. The kid who occasionally says “wait, I think the answer is actually different” has been building their own understanding alongside their AI use. That is a real distinction worth encouraging, not just an inconvenient habit of questioning the tool.

The most useful thing you can do this week is introduce one low-stakes habit: after any AI-assisted homework, ask one follow-up question about what they just turned in.

Pick one thing from the assignment and ask your kid to explain it to you. No notes, no AI. Just you asking a genuine question about what they just did. If they can answer easily, AI helped them learn. If they go blank, that is useful information for both of you. Treat it as data, not judgment. “Let’s figure that one out together” lands better than “you should not have used AI.” The goal is to calibrate, not punish.

Model the check yourself. Tell your kids about a time you read something online or got an AI answer and then forgot it within a day. Or a time you had to sit with something three times before it really clicked. Learning Honesty is not a lesson you deliver once. It is something your kids watch you practice. If you treat your own learning honestly, they have a template for what that looks like in a real adult.

Occasionally ask at dinner: what did you learn this week that you could explain to someone else? Not as homework review. Just as a genuine question. The habit of being asked “what do you actually understand about that” builds a more accurate internal monitor for comprehension. Kids who grow up with that question develop an earlier, more reliable sense of when something has landed and when it has not, which is the whole skill.

The survey found that most teens are using AI without guidance. Most teachers are worried but working within real constraints. The gap between what students hand in and what they actually know is growing. Your kid does not need to stop using AI. They need one habit that keeps their own understanding in the conversation.

My oldest will keep using AI. I am not going to fight that. What I want is for her to eventually hold it the way she holds any tool she relies on: with a clear and honest sense of what she knows because of it and what she still needs to work through herself. That distinction is the one worth teaching before the school year sets the pattern for how she treats every assignment.

The Common Sense Media survey found that 40 percent of teens already sense they are missing something when they use AI for school. That awareness is a starting point. Your job is to give it a name and a habit to go with it. The name is Learning Honesty. The habit is one question: Can I explain this without AI? If your kid can answer yes most of the time, they are learning. If not, you now know exactly where to start.