How To Teach Kids To Check If AI Information Is Accurate: Tips From A Middle School Teacher
Last month, I sat in the back of my 7th grade life science class listening to Javi present his project on the Karner blue butterfly, an endangered species that lives in the oak savannas an hour north of our Wisconsin town. The sun was hitting the peeling blue paint by my classroom radiator, and most of the kids were slouched in bean bag chairs, half paying attention, until Javi said Karner blue caterpillars only eat poison ivy. A few kids nodded—poison ivy is everywhere around our local hiking trails, so it sounded reasonable. I didn’t interrupt him mid-presentation, but when he finished, I jotted a note to talk about it after class.
When I asked Javi where he got that fact, he shrugged and said he’d asked a popular AI chatbot. His local library didn’t have any recent books on local butterflies, and the first few Google links were behind nature organization paywalls, so AI gave him a fast answer. It sounded right, he said, so he didn’t think to check anywhere else. That moment wasn’t a surprise to me. As a teacher and a mom of a 4th grader, I’ve seen how comfortable kids are using AI for everything from homework help to cookie recipes, but almost no one has shown them how to make sure what AI says is actually correct. Last week, my own daughter almost turned in a history report that said Susan B. Anthony died in 1905, a full year off the actual date, because that’s what her chatbot told her.
I used to think this was just a kid problem—they’re too trusting, they don’t want to do extra work. But after a few months of talking through this with my class and my kid at home, I’ve realized it’s just a new skill we have to teach, like learning to check a map or proofread a paper. It doesn’t require a big unit or a bunch of extra homework, just small, repeated habits that stick.
After Javi’s presentation, we spent 15 minutes the next day walking through what happened. AI had mixed up Karner blue caterpillars with another local butterfly that does feed on poison ivy, a common mistake that’s easy to make if you don’t have the specific context right. Then I showed the class the three small steps I use any time I get information from AI, steps they can use for homework, projects, even that cookie recipe they’re pulling up for a 4-H bake sale.
First, do a 10-second gut check before you write the fact down. I tell kids to ask themselves two simple things: Does this surprise me? Does it line up with what I already know about this topic? If the answer is no to either, that’s a red flag, not something to just accept because it sounds confident. Javi told me later he’d thought it was a little weird that a rare endangered caterpillar would eat such a common plant, but he didn’t stop to question it because AI stated it so clearly. Now he marks those little niggling doubts to check, instead of brushing them off.
Second, cross-check any key fact with at least two non-AI trusted sources. I don’t make kids hunt all over the internet for these. I have a list of 12 kid-friendly, ad-free, paywall-free trusted sources pinned to our Google Classroom: things like our state DNR website for local nature facts, National Geographic Kids for science, the Library of Congress for history, and our state university’s extension page for everything from gardening to food recipes. At home, I saved that same list to my daughter’s bookmark bar on her laptop, so she doesn’t have to remember what to use or sort through random, unreliable blog posts. For Javi’s butterfly fact, we pulled up the Wisconsin DNR’s Karner blue page and the United States Fish and Wildlife Service page side by side, and both clearly said caterpillars only eat wild lupine. It took less than a minute to confirm the mistake.
Third, if you used AI to write a whole draft or outline, go through line by line and pull out every factual claim before you turn it in. Numbers, dates, species names, locations—those are the things AI mixes up most often, so those get a check. One of my students, Mia, told the class she once followed an AI chocolate chip cookie recipe that called for three tablespoons of salt instead of three teaspoons, and the whole batch was inedible. After that, she checks all measurement numbers twice, which is the same habit she uses for school facts.
I don’t ban AI in my class, and I don’t tell kids it’s all wrong. It’s a useful tool, I tell them, just like a textbook or a friend who helps you study, and all of those can get things wrong too. It just needs a second check.
Some days it works, some days it doesn’t. Last week, a kid turned in a project that said there are no wild mountain lions left in Wisconsin, which is wrong—we’ve had confirmed breeding populations for years now. He’d forgotten to cross-check, so we just walked through the steps again together. It’s not a skill you teach once and it sticks forever. It’s something you practice, over and over, every time a kid pulls up a chatbot to answer a question. For now, that’s enough. More kids are stopping to check their facts than they were six months ago, and that’s all I can ask for right now.
Please indicate: Thinking In Educating » How To Teach Kids To Check If AI Information Is Accurate: Tips From A Middle School Teacher