子供たちが語るAIへの実感：Cheatingや深層偽造への懸念など

When we set out to talk to kids about artificial intelligence, we thought we knew what we’d hear. We expected some to tell us they were using it to cheat a little, the way Millennials and Gen Xers opened up CliffsNotes or programmed formulas into their TI-82s, and others to share inspiring ways they were using it. We were also listening for concerns that were less kid-specific, like deepfakes or job destruction. But what we actually heard when we asked kids aged 10 to 18 about AI had tons of nuance.

Many of the same kids who can go on and on about music, rock climbing, or soccer met our questions with words like “bruh” and “meh”—or were so deeply against AI or uninterested in making it part of their lives that they didn’t want to talk about it at all. One teen said his peers use it for things they know they shouldn’t, like writing papers. One told us she won’t touch AI because of the environmental impact. A few said they find the whole field disheartening: “AI isn’t the solution to our problems,” said Winter, a 17-year-old. “I’m afraid it’s going to be the end of creativity and critical thinking.” Yet most of the kids we asked admitted to using AI at least a little bit.

AI doesn’t yet seem to be something a lot of elementary- or middle-school-age kids we spoke to are focused on—and they aren’t begging for it, the way they do for iPhones and Snapchat accounts. Many told us that some of their first AI encounters came from their parents or schools. Sometimes, they said, it’s just embedded in the devices and apps they already rely on. It’s just there, in things like a Google search.

What we heard tracks with the data. In a survey published in February 2026, the Pew Research Center found that 57% of teens in the US had used chatbots to search for information, 54% tapped them to help with schoolwork, and 47% had used them for fun or entertainment. Only 12% had used them for emotional support or advice. Teens are over four times more likely to be using AI in innocuous ways than potentially harmful ones. Some are even using it to build things, whether it’s a character, a tech platform, or a tutor to help other kids study.

None of that means the worries are misplaced. Kids can stumble into unfiltered content, lean on a chatbot instead of their own judgment, or trust an answer that’s wrong—and they should be protected from those dangers. But the danger is the reason to teach the thing, not to avoid it. We don’t teach teens to never drive. We teach them to check their blind spots.

What surprised us most was how much young people might be able to teach adults about AI, and how clearly the kids who use it could name what they will and won’t hand over. They’re not as worried that it will take their jobs as they are that it might impact society. And, with increasing access to tools that could in theory do their thinking, their talking, or even their friend-­making for them, it sounds as if most want to keep their hands on the wheel.

Interviews have been edited for length and clarity.

imageJUSTYNA STASIK

The Coder

Remy, 16, New York

The word that comes to mind when I think about AI is “indifferent.” I just don’t find the current applications that exciting for my own use. I go to school. I teach tae kwon do. I read. I play games with friends. None of that needs AI. I mean, I use it. I mostly use Claude, the free version, for programming outside of school. I had it help me write a program to see if I could tweak my computer’s overclock. So I see the appeal.

But at school, I actually think AI mostly makes my assignments worse, not better. In English, everything is now in-class writing, because teachers don’t want kids cheating. So we have only 70 minutes to write a whole essay, and I think that hinders my writing. (Did you know Princeton voted to let faculty proctor exams for the first time in over a century? Their honor code goes back to 1893, and now it’s over because of AI.)

As far as code goes, I’d also rather build things myself. I’ve been making a reinforcement-learning model in a game engine with a friend; it moves randomly at first, gets rewarded for walking toward a coin, and after enough iterations it teaches itself the most efficient path. I’ve also tested AI for game development, and it isn’t there. It makes sloppy code, and it’s bad at blending mechanics into something cohesive. I’d spend more time correcting it than writing it myself.

I think AI right now is sort of like the first car or the first airplane. It’s interesting but crude. It’s obviously an amazing invention but not actually good yet.

Overall, I think AI right now is sort of like the first car or the first airplane. It’s interesting but crude. It’s obviously an amazing invention but not actually good yet. It’ll get somewhere. One thing I read about was AI flagging breast cancer more accurately, trained to catch its own false positives so a human still verifies. That’s the version I care about.

The Organizer

Danielle, 18, California

I’m studying engineering, and my life goal is to innovate technology that will help as many people as I can. The way I see it, AI isn’t inherently good or bad; that’s decided by the people using it. It’s already being used for lots of good. Just think about how it helps people with personalized education and more accessible medical diagnoses.

imagePING ZHU

So far, the biggest project I’ve worked on with AI is called Next Voters. My teammates are more on the technical side, and I’m working on scaling. Right now, we’re focusing mostly on city councils as well as states. Our system turns dense, hundred-page documents into a headline and a few plain-language bullet points in your inbox. And everything is cited, so you can click straight to the actual policy to learn more. The information comes to you, instead of you having to remember to go search or prompt for it.

One AI agent finds official government sources for a given city or state—the council website, the proposed bills, the meeting transcripts. Another verifies they’re real and credible; another scrapes them every week for the latest updates; another sorts them into categories like civil rights, immigration, and economics; and the last one writes our weekly newsletter.

The project’s goal is to reduce the barriers to democratic participation—to make sure anyone, regardless of race, gender, income, or education level, has an easy way to get the information they need and then think critically about how they want to use it. We made it because right now, it feels as if most teens aren’t very engaged. I was in English class when the war in Ukraine came up and someone said, “There’s a war going on?” That gap, plus all the emotionally charged social media misinformation that gets promoted because it earns the most clicks, makes me nervous for the next generation of voters.

We don’t want AI to think for people; we want to use it to disperse knowledge. In other words, we want to deal people the cards and let them play them however they want, but we have to make sure they have the cards in the first place.

The Cringe-o-meter

We asked kids to rate a range of AI uses from totally fine to not okay.

imageCHRIS PIASCIK

The Naturalist

Hazel, 17, New York

When ChatGPT first came out, my dad showed it to me and it seemed fun. But as it got more prominent and seemed to be everywhere, I started to feel uneasy. Then I learned about the environmental impact.

I’m a rock climber and I hike a lot. It’s good because when I’m on a wall, I’m just focused on staying on that wall. I’m not thinking about my phone or anything else. That’s why I love it. I also love the views and being around animals—even insects. I want to be an ecologist, and the more time I spend in nature, the more I want to protect those wild spaces.

The part that bothers me most about AI is the data centers that companies are building to enable it. They house these huge blocks of servers that use enormous amounts of water. They take it from local towns and don’t leave enough behind for the people who actually live there. And when they get big enough, they put off so much heat they can raise the local temperature a degree or two.

So I make small choices. When AI pops up somewhere, I just don’t engage with it. It can feel isolating when everyone around me is using it, but I don’t want AI to be the thing that kills the places I love.

imagePING ZHU

The Storyteller

Wesley, 14, Ohio

My friends and I have all heard about AI and seen videos made by AI, but I mostly use it for school. I wrote a short story and ran it through ChatGPT to catch my grammar and spelling errors, and I used it to debug a little game I’d coded for a project. What I worry about is it robbing us of our ability to think creatively, or to think for ourselves.

But I have tried using AI for fun. When I was bored, I tried to have a conversation with ChatGPT once or twice, but I didn’t really like it. Character.AI is more fun. You type in all this information, give it a bunch of prompts and a profile picture, and then you can post your AI character for anyone to use. You just put what you’ve made out there. Then you talk to it. My favorite show is One Piece on Netflix, so I threw myself onto its pirate crew using a character I found.

Other people have used Character.AI to build whole games. There’s a rap-star simulator where you pick your difficulty and where you’re from, and the AI creates a game out of that. There are also World War II simulators, and chats where you’re working with assassins from a TV show. You can find pretty much anything.

I guess I’d recommend it, but with caution. The content is pretty unfiltered, so you have to be careful what you click on. You learn its limits fast, too. On the free version the memory runs out: Get far enough into a chat and it slows down and forgets what happened. It’s like everything else with AI. If you trust it to run on its own, it falls apart. You have to keep steering it where you want it to go.

I guess I’d recommend it, but with caution. The content is pretty unfiltered, so you have to be careful what you click on. You learn its limits fast, too.

imageJUSTYNA STASIK

The Artist

Sylvia, 10, Michigan

I haven’t used tools like ChatGPT or Claude myself, but my mom does. I really like to draw and write songs, but I don’t use AI for that. I don’t really have big feelings about AI either way. It’s a little like a calculator. A calculator does the math for you, and AI does other things for you. But I don’t like when AI tricks you, like when my mom found some songs she liked on Spotify and then looked up the artist to see what they looked like. It turns out the whole thing was made by AI. I was surprised, even though I still like the song.

I do use AI at school, through a program called SchoolAI. Mostly I put my writing in and it gives me ideas or helps me revise. You can’t have it just write for you, but you can use it to help. When I’m older I want to be an artist, or maybe a librarian. I’d probably use some technology either way. But the drawing and the songwriting? Those I want to keep doing myself.

The Cringe-o-meter (continued)

We asked kids to rate a range of AI uses from totally fine to not okay.

imagePING ZHU

The Pre-Premed

Evelyn, 13, Oregon

In January, I was diagnosed with type 1 diabetes, and that’s when AI became a bigger part of my life. Now when we’re cooking, we can run a recipe through ChatGPT, tell it the serving size, and it works out how many carbs there are. We use AI like that a lot.

My glucose monitor and my insulin pump also talk to each other using their own kind of AI to predict dosing. The monitor tracks what my blood sugar actually is, and the pump does the math. So if it predicts that my blood sugar will be high in 30 minutes, it gives me a correction dose, and if it predicts I’m about to go low, it stops the insulin before that happens. When I was first diagnosed I was still doing shots, and I went low almost every night. It was really stressful. Now the pump can catch it, and at night my phone goes off if I drop, so I wake up and drink juice. Mostly, I just get to sleep more because of it.

But the hardest part of having diabetes isn’t something I can use AI for. It’s remembering to carry all my supplies everywhere—to school, to a long day of anything.

I do use AI for school sometimes. Memory tricks when I’m studying for a test, ideas to get a project started. It’s a really good tool for that. But I don’t know exactly how I’ll use AI in the future. I want to be an endocrinologist someday, so I figure something will come up, since I’m already using it to help with my diabetes. I know other people worry that AI is going to take over the world. I don’t really think so. I still think we’re in control of it, and I think the benefits outweigh the risks.

I don’t know how exactly I’ll use AI in the future. I want to be an endocrinologist someday, so I figure something will come up, since I’m already using it to help with my diabetes.

The Inventor

Krishiv, 17, Ontario, Canada

When I was growing up, I always liked building things: Lego builds, Minecraft worlds, and then video games in Scratch. I’d make a little game, post it for other kids to play, read the comments, and make it better. Then, when I started high school, I had to spend way more time studying than I ever had, and honestly I just wanted to build things. So I went looking for ways to get good grades while studying less. Khan Academy had an AI tutor in the works, but it was stuck behind a waitlist, so I figured, why not build my own?

After months of launching random stuff, I created an AI tutor called Aceflow. You could feed it anything a teacher assigned—a 30-minute lecture video on YouTube, a blog post, a PDF of the textbook or presentation slides—and it would spin up endless practice questions, with a tutor on the side that explained things the way my teacher did. I built it just for myself, showed it to my friends, then put it on TikTok. It got tons of views on TikTok and thousands of users.

imageJUSTYNA STASIK

Was I worried people would call it cheating? Not really. I knew how to defend it: A tool like this isn’t so different from well-off families hiring expensive private tutors, except everyone gets one. That part mattered to me. Back in eighth grade, a teacher had me run a little computer science class for about 30 kids with special needs, and once they got personalized attention, they were building games nobody expected of them. That convinced me that kids are capable of so much more than people think, and AI can help scale that level of personalized attention to everyone. That unlocks so much potential.

That first AI tutoring project ended up helping me land part-time roles at BenchSci (one of Canada’s biggest AI companies) and Simple Ventures (a venture firm). More recently, I joined an AI lab at MIT; co-instructed an AI agents course with an MIT professor; and launched CheetahPrep.com, an SAT prep platform that uses AI to adapt to each student.

I’m generally optimistic about how AI will impact humanity, but when other kids’ first reaction is fear, I think that’s an important sign too. It’s a reminder that we should be excited about the future while still being mindful of the risks, working together to make AI work for humanity.

Jen Swetzoff and Keeley McNamara are the founding editors of Anyway, an independent print magazine for tweens and teens.