Why Kids Lie About AI Doing Their Work: Lessons From ‘The Machine that Cried Robot’
The gym smelled like hot glue and sour lemonade at last month’s 4th grade maker fair, and I was balancing a clipboard, a half-eaten cookie, and a stack of participation ribbons when I stopped at Milo’s table. His project was a Lego sorting machine, built from a pile of old bricks and a $10 sensor he’d asked his grandma for for his birthday. It hummed soft when I leaned over, and when he dropped a red brick on the chute, it slid right into the red bin. I asked him to walk me through how he’d calibrated the sensor to tell colors apart, and he went bright red, crossed his arms over his hoodie, and blurted, “AI did it. ChatGPT wrote the code. I didn’t do anything.”
I’d seen Milo working on this thing every recess for three weeks. I’d walked past him on Tuesday, hunched over his laptop in the corner, muttering at the screen and crossing out lines of code so hard the tip of his pencil broke. I knew he’d done the work. This wasn’t the usual “I let my friend do it” confession I’ve come to expect at project fairs. This was the opposite: he was taking credit away from himself, handing it off to a robot. I thought of the boy who cried wolf, flipped on its head: the machine that cried robot.
I sat down on the cold gym floor next to his table, because all the folding chairs were taken by parents, and asked if I could see the code. He pulled up his Google Drive file, and there it was: 27 lines of code, with his messy handwriting in the comment boxes, crossed out variables, and a note scrawled right in the middle of line 12: stupid sensor is dumb. ChatGPT doesn’t write that. I pointed to the note, and said “You wrote all this, didn’t you?” He stared at the scuff on his sneakers and didn’t say anything for a minute.
After the fair, we walked back to my classroom, I grabbed two granola bars from the drawer I keep for kids who forget lunch, and sat with him at the back table. He finally told me what was going on. His older brother, a high schooler, had been over for family dinner the Sunday before, and heard Milo talking about his coding project. He’d laughed and said any idiot could do that themselves, only dumb kids don’t use AI to do their work now. His dad had joked along, saying pretty soon AI will do every job anyway, so why bother learning to code from scratch. Milo said he got nervous when I asked him about the code, and suddenly he was sure if he said he did it himself, I’d think he was slow, or his friends would make fun of him.
Over the past three years of AI being part of every conversation in schools, I’ve sat through more training sessions on how to catch kids cheating with AI than I can count. We have new software, new rules, new checklists to run AI detection on every essay and project. What no one talks about are the kids like Milo: the ones who lie and say AI did their work, when they did every bit of it themselves. I’ve had three of these moments in the last three months alone: a fifth grader who told me AI wrote her personal narrative about her dog, when she’d stayed after school twice to revise it, a third grader who claimed DALL-E drew his dinosaur when he’d spent two weeks working on it for art class.
It’s easy to brush these off as just kids being weird, or wanting to fit in with older kids. But the pattern here is clear: kids are listening to every big scary conversation we have about AI. They hear us talking about how AI is going to take all our jobs, how kids need to learn to use AI to get by, how any work that doesn’t use AI is outdated. They absorb the idea that AI is better, faster, more impressive than anything they can do themselves. So when it’s time to show off their work, they panic, and give all the credit to the robot.
After Milo’s project, I changed a few small things in how I talk about AI with my class, and I’ve shared these small shifts with parents at conferences that work just as well at home. First, I stopped opening any conversation about student work with “Did AI do this?” That question immediately assumes cheating, and puts kids on the defensive. Instead, I now ask “What part of this did you work on the most?” It’s a small change in wording, but it gives kids room to talk. If they used AI for a part of it, they can say that. If they lied about AI doing it all, it gives them a low-pressure way to backtrack without feeling like they’re in trouble.
Second, I started talking about AI like what it is: a tool, not a replacement for the kid. A lot of kids only hear that AI does everything, so I normalize talking about my own use of it out loud. I’ll say to the class, “I asked AI to give me a list of good read aloud books about space, but I picked the one I thought you’d like best, that’s my call.” For parents, this works just as well at home: if you use AI to plan a birthday party menu, say out loud “AI gave me these ideas, but I’m taking out the broccoli because no one here likes broccoli, that’s my change.” Small, everyday moments that make it clear AI doesn’t do the actual work of living or creating, we do.
Third, when a kid does admit they lied about AI doing their work, I don’t call them out in front of other people, and I don’t lecture them about lying. I just point out the specific things they did that AI couldn’t do. For Milo, I said “You thought of this machine because you’re always losing your blue Lego bricks, right? AI doesn’t know you lose your blue bricks. You adjusted the height of the chute three times because your cat knocked it over, that’s your work.” Specific, concrete, not vague praise. It helps them see their own work is worth claiming.
Milo ended up getting second place at the maker fair, and he taped that “stupid sensor is dumb” note right to the front of his machine when he took it home. I still get these moments, though. Last week, a girl turned in a poem about her grandma’s garden and wrote “AI helped” at the top, even though all she did was ask AI for a list of rhymes for “rain.” She just thought she had to say AI helped for it to be good. We don’t have all the answers for what AI means for kids and school yet. But right now, the most important thing I’ve learned is that we spend so much time worrying about kids letting AI do their work, we miss the real problem: kids are already convinced their own work isn’t good enough. Shifting the question from “did AI do this?” to “what did you do?” isn’t a fix for everything, but it’s a start.
Please indicate: Thinking In Educating » Why Kids Lie About AI Doing Their Work: Lessons From ‘The Machine that Cried Robot’