How to Teach Kids to Check If AI Information Is Accurate: A Teacher’s Simple Tips
Last Wednesday, I was wandering between rows of poster boards in my 7th grade digital literacy class, stepping around scattered glue sticks and crumpled blue construction paper, when Javi waved me over. His project was on native Vermont pollinators, and his big hand-drawn blueberry bee took up half the poster. “AI says these only live in Maine, so I wrote we don’t have any here,” he said, wiping a smudge of glue off his mud-streaked jeans—leftover from riding his bike to school that morning.
I didn’t want to just tell him he was wrong. I’d done that enough times before, and all it taught kids was that AI gets me in trouble, not that they need to check the work themselves. Before I could say anything, Mia, who sat two rows over, leaned over her own poster on milkweed and said, “Wait, I helped my grandma tag bees in her blueberry patch outside Stowe last summer. I definitely saw those. I have a photo on my phone.”
We gathered the class around our jittery old portable classroom projector, and Javi pulled up the conversation he’d had with AI. His first question was simple: “Where did you get the information that blueberry bees only live in Maine?” AI responded that it drew from multiple range maps and observational data, but couldn’t link any of it or name a specific source. That was our first red flag for the day.
Next, I walked the class through pulling up the USDA’s free public pollinator range map. It took 20 seconds to pull up, and sure enough, blueberry bees were clearly documented across all of New England, including most of Vermont. We clicked over to the University of Vermont extension service page for home gardeners, and it listed blueberry bees as a common local pollinator for backyard blueberry bushes. Two independent sources, both from organizations with a clear reason to get that fact right. Javi erased the line about them only living in Maine and wrote the new info in. That was the end of the poster project, but it started me thinking more about how we teach this skill, not just in my class, but at home too.
A week later, at back-to-school night, a mom named Lara pulled me aside to talk about her 5th grader. He uses AI to help with homework all the time, she said, and last month he turned in a book report that said the main dog in Where the Red Fern Grows was named Buck. Buck is the dog from Call of the Wild, of course. Lara had told him he should have checked, but he just shrugged and said “AI said it, I thought it was right.” She didn’t know what to tell him beyond “don’t trust AI,” and that didn’t seem to stick.
I’ve found that warning kids to not trust AI is as useless as telling them to not talk to strangers. It’s too vague, and they still need to use the tool for school and other things. Instead, I teach three quick, low-effort questions that kids can ask every time they get a fact from AI, and parents can use these at home too, no extra tech skills or lesson plans required.
The first question is: Does this match what I already know? In Lara’s case, her son had just finished reading the book, so he could have stopped for 10 seconds and asked himself that. He would have remembered the dogs were named Old Dan and Little Ann, and that would have popped up as a red flag right away. For Javi, Mia’s personal experience seeing blueberry bees locally was the first check. Kids don’t need to look anything up for this step—they just need to pause and compare the AI answer to what they already know from class, reading, or their own life.
The second question is: Can I find this same fact in two other places that aren’t AI summaries? It doesn’t have to be fancy. For most school projects, that means sticking to sites that don’t publish AI-generated content full time: government agency pages, university extensions, museum sites, major kids’ news outlets, or their own textbook. It doesn’t take more than five minutes to cross check, and two matching sources from independent groups are almost always enough to confirm a fact is right.
The third question is: Can AI show me where it got this information? If AI can’t cite a specific source, that’s automatic reason to double check. AI doesn’t pull facts from a library of verified information—it puts words together based on patterns, so it’s very good at making up facts that sound real. If it can’t tell you where the info came from, you can’t confirm it’s right.
This isn’t about banning AI. AI is a tool, just like the encyclopedias I used as a kid that had wrong facts in them sometimes, or even the textbooks we use in class that have typos and outdated stats. The point isn’t to avoid the tool, it’s to learn how to check it. I had a student last year who used AI to outline his paper on local sea level rise, then checked every stat AI gave him against NOAA data, adjusted the wrong estimates, and turned in one of the best papers in the class. That’s the outcome we want, not no AI use.
It’s not a skill that sticks after one lesson, of course. Last Friday, I got a history paper from an 8th grader that said Susan B. Anthony voted in the 1920 presidential election, after the 19th amendment passed. AI had mixed up Anthony’s 1906 death date with the amendment’s ratification. But an hour after I got the paper, the kid emailed me an updated version. He’d been reading over his work before the final turn-in, remembered we’d talked about Anthony the semester before, asked himself the three questions, checked the National Women’s History Museum site, and fixed it.
We still have days where half the class forgets to check and turns in wrong facts. It’s slow going, and I don’t expect every kid to remember every time. But more and more often, I see them pause, pull up an extra source, and catch the mistakes themselves. That’s enough for right now. It’s a work in progress, just like most skills kids learn.
Please indicate: Thinking In Educating » How to Teach Kids to Check If AI Information Is Accurate: A Teacher’s Simple Tips