Does AI Dependency Change How Kids Approach Hard School Problems? A Teacher’s Observations
Last Tuesday, I stayed after school to help my 8th grade math students with their systems of equations word problems. Mia, who always wears mismatched fuzzy socks and keeps a half-melted cherry slushie in the cupholder of her desk, was stuck on the last problem about farmers planting corn and soybeans. Before she even wrote down what the question was asking, her hand drifted to her phone in her hoodie pocket.
I leaned against her desk and said, “Hold up. What do you know so far?” She blinked at the paper, then back at me. “Nothing. It’s too confusing. I’ll just have AI give me the steps.” That’s when I really noticed the shift I’ve been tracking for the past 18 months. Two years ago, before ChatGPT became a household name for middle schoolers, Mia would have grumbled, crossed out two wrong attempts, drawn a messy little diagram of the farm to make it make sense, and asked me a targeted question after five minutes of trying. Now, reaching for AI is the first move, not the last.
A month earlier, I watched my 7th grade science groups build model bridges out of popsicle sticks. Every group pulled up AI first to get the “best” design before they touched any materials. When I started teaching 12 years ago, kids would cut three wrong designs, glue them together crookedly, watch one collapse halfway through weight testing, and only then look up how to fix it. That mess was where the learning happened. Now, the mess is something they avoid before it even starts.
It’s not that all kids are dependent on AI to do their work, and it’s not that AI doesn’t have a useful place. I use it to make extra practice worksheets and generate varied example problems for struggling students. The change I see isn’t even mostly about cheating. It’s about how kids approach difficulty. A hard problem used to mean, okay, I need to sit with this for a few minutes, try a few wrong things, and slowly work it out. Now, a hard problem means, why am I even struggling when I can get a clear answer in 10 seconds? That default shift is the real change, the one that affects how kids learn to solve problems long term.
I saw this play out at my own kitchen table two months ago, when my 10-year-old son was working on a narrative writing prompt about a time he messed up. I walked by to grab a mug of tea and saw he’d already pulled up AI to generate a story about forgetting his soccer cleats for the big game. He hadn’t even jotted down one idea of his own. I didn’t take his phone away or yell about cheating. Instead, I asked him to do one quick thing first: write down three of his own bad memories that he actually thought were funny. He wrote down the time he dyed our cat blue with food coloring, the time he burned his grandma’s birthday cake, and the time he accidentally called his principal “mom” over the school intercom. We looked at AI’s generic story next to his ideas, and he decided his cat story was way more interesting. That’s when it clicked for me: we don’t have to ban AI. We just have to put one small, intentional step between the kid and the shortcut.
I started testing a simple rule in my classroom that a lot of my parent friends now use at home, too. It’s the 3-step check, no fancy training or strict punishments needed. Before anyone pulls out AI for a hard problem, they have to write down three quick things: what the question is asking, what they already know that might help, and one guess at the first step to try. That takes two minutes, tops, but it forces kids to stop and engage with the problem before they hand it off to someone else. For parents at homework time, another small shift that works: instead of asking “do you need help with that?” as soon as they frown at their homework, ask “what’s the first thing you’re going to try?” That tiny question shifts their mindset from looking for an out to looking for a way forward. If they do end up using AI to check their work or get an explanation, ask them to do one more quick thing: explain the second step of AI’s answer in their own words. That way, they’re still doing the thinking, not just copying the output.
I don’t pretend this solves everything. Some kids still sneak AI to do their whole homework, some forget the rule and reach for their phone before I can blink. It’s also true that some kids who struggle with anxiety around school work use AI to avoid the stress of feeling stupid, and that’s a conversation we have to have one on one, not just with a hard rule.
Mia still reaches for her pocket first most days, but now she usually catches herself, huffs a little, pulls her scratch paper closer, and starts writing. Two weeks ago, she solved that corn and soybeans problem all on her own before she even touched her phone, and she held up the paper grinning like she’d just won a prize. Then she asked AI to check it, just to be sure, and that’s okay with me.
We’re all still figuring this out. AI isn’t going away, and it’s not the end of learning. The hard part is just remembering that struggling for a few minutes isn’t a waste of time. It’s how you get good at the hard stuff. That’s the lesson we’re still teaching, one small problem at a time.
Please indicate: Thinking In Educating » Does AI Dependency Change How Kids Approach Hard School Problems