Last updated: April 2026
My eight-year-old asked ChatGPT how many bones a shark has. It gave him a confident, well-formatted answer: 206. That’s the number of bones in the human body. Sharks have zero bones. Their skeletons are made of cartilage. That thirty-second exchange taught me more about kids’ critical thinking AI than any parenting book I’ve read.
He didn’t flinch. He copied it straight into his notes. Why would he question it? The answer looked clean, sounded sure, and came back in two seconds. That moment taught me something I’ve been thinking about ever since: kids’ critical-thinking AI habits don’t develop on their own. Not when the tool is designed to sound right even when it’s wrong.
So here’s the one question I now make my son ask before he trusts any AI-generated answer:
“How would I check this if the AI didn’t exist?”
That’s it. Not complicated. Not technical. But it changes the way a kid relates to the machine entirely.
Why This One Question Builds Kids Critical Thinking AI Skills
Most advice about kids and AI boils down to “teach them to think critically.” Fine. But that’s like telling someone to “be healthier”, it’s too vague to act on, especially for an eight-year-old.
“How would I check this if the AI didn’t exist?” works because it does something concrete. It forces the kid out of the AI’s frame and into their own. They have to think about where real information lives.
For the shark question, my son thought about it and said, “I guess I’d look in my animal encyclopedia.” He pulled it off the shelf. Thirty seconds later, he found the answer: zero bones, all cartilage, and gave ChatGPT a look I can only describe as personal betrayal.
A kid who catches the AI being wrong once starts reading every answer a little differently.
What AI Gets Wrong (and Why Kids Don’t Notice)
AI language models don’t know things the way a person does. They predict what word comes next based on patterns. That means they’re excellent at sounding authoritative and terrible at knowing when they’ve crossed from fact into fiction.
Adults miss this, too. I have for sure. But kids are especially vulnerable for a few reasons:
– They haven’t built up enough background knowledge to spot errors.
– They’re used to trusting tools that adults hand them.
– The answers look like textbook entries, numbered, formatted, and polished.
Imagine a ten-year-old researching a school report on the solar system. She asks an AI tool how far Neptune is from the sun. It gives her a number. Maybe the number’s right. Maybe it’s off by a billion miles. She has no reference point. Without a habit of checking, that wrong number goes into the report, gets turned in, and gets reinforced.
The question, “How would I check this if the AI didn’t exist?”, builds that reference point over time. A kid who keeps asking it starts to learn where reliable information actually lives: a book, a verified website, a parent, a teacher, a primary source.
Making It a Habit, Not a Lecture
Here’s where most parents (myself included) get it wrong at first. You can’t just explain this once and expect it to stick. Kids don’t learn habits from speeches. They learn habits from repetition in real moments.
What’s worked in our house and in classes at Livingston Global Academy is simple: make the question part of the workflow, not a punishment after a mistake.
When a kid uses AI for a project, the checking step is built in. Not “you got it wrong, go verify.” More like, “Cool, now where’s your backup source?” It’s the same energy as “show your work” in math. It’s just expected.
One of our students, the one who sells cookies baked from his grandpa’s recipe, was researching food safety guidelines for a class project on his business. He asked an AI assistant about safe baking temperatures. The answer was close but not quite right for the specific item he was making. Because he’d already gotten into the habit of checking, he pulled up the USDA’s actual page and caught the discrepancy himself. No adult had to flag it.
That’s the goal. A kid who checks automatically, not because someone’s standing over their shoulder.
The Bigger Skill Behind the Question
This isn’t really about AI. Not entirely. Kids’ critical thinking AI habits are the surface, the real skill underneath is older and more durable.
A kid who learns to ask “how would I verify this?” at age nine is building something that lasts well past any single technology. They’re learning to separate confidence from correctness. Even adults still struggle with it in the news, in advertising, in conversations.
AI just makes it urgent. Because AI is confident every single time. It doesn’t hedge. It doesn’t say “I’m not sure.” It delivers wrong answers with the same polish as right ones. For a kid growing up inside that environment, the ability to pause and check is the difference between being informed and being misinformed by a very convincing machine.
The parents I talk to often worry about the wrong thing. They worry about screen time, about whether AI will make their kids lazy. Those are real concerns. But the sharper risk is a kid who trusts without verifying, because that habit, once set, is hard to break.
How to Start This Week
You don’t need a curriculum. You need a Tuesday afternoon.
Next time your kid uses an AI tool for homework, a question, or just curiosity, try this:
1. Let them get the answer.
2. Ask: “How would you check this if the AI didn’t exist?”
3. Let them figure it out. A book, a website, asking you, anything counts.
4. Talk about what they found. Did the AI get it right? Close but wrong? Completely off?
Do that five times, and you’ll notice something shift. They’ll start checking without being asked. They’ll start wanting to catch the AI slipping up. It becomes a kind of game, and kids are competitive enough to enjoy winning.
Questions Parents Ask
Age eight is a practical starting point. Kids that age can compare two sources, an AI answer and a book or website, and spot basic differences. Younger children can do it with help, but independent verification tends to click around age 8 or 9.
ChatGPT, Claude (with a parent present), and Google’s Gemini all work for supervised practice. The specific tool matters less than the habit. Any AI chatbot that gives text-based answers will produce enough errors for a kid to catch.
No, the opposite tends to happen. Kids who learn to verify become more confident in using AI because they trust their ability to distinguish good output from bad. It’s the difference between using a tool and being used by one.
It depends on the topic, but a 2023 study from Purdue University found that ChatGPT gave incorrect answers to programming questions over 50% of the time (It’s much better today in 2026),and 77% of those wrong answers were unnecessarily verbose, which made them seem more credible. Non-technical topics vary, but the pattern holds: confident delivery, inconsistent accuracy.
Both. Regular practice at home builds a daily habit. A structured class adds depth, kids learn to evaluate AI-generated content together, debate when the AI might be wrong, and practice verification in real time with a teacher guiding the conversation. The combination is stronger than either one alone
If you’re looking for a place to start, Livingston Global Academy offers live online classes where kids ages 8–12 learn AI literacy, critical thinking, and real-world skills together. No coding experience required, just curiosity.
Andrew Livingston is the founder of Livingston Global Academy and writes about AI, education, and raising kids who think for themselves in an AI-saturated world.