What Happens to Kids’ Cognitive Resilience When AI Eliminates Productive Discomfort?
Last Tuesday, I sat across from my 7th grade tutoring student Mia at the kitchen table, a half-eaten bag of cheddar goldfish spilled between us, while she stared at a page of slope word problems. Before she finished reading the first question, her thumb was already unlocking her phone, and the AI chat app popped up. I didn’t call it cheating. I just asked, “Can you show me what you think the question is asking first?” She shrugged, locked her phone, and said, “What’s the point? This is faster, and it’s never wrong.”
I thought back to when I was 12, sitting at my own kitchen table stuck on a nearly identical problem. I erased so many times I put a hole in my graph paper, and I still remember the jolt I got when I finally connected the speed of the car in the word problem to the slope of the line. That moment of confusion, that uncomfortable, itchy “I don’t get it” feeling stuck with me for 26 years, because I worked through it. Right now, all anyone talks about with AI in school is cheating. Parents panic, schools ban phones, teachers patrol aisles checking screens. But that whole debate misses the quiet, bigger problem happening right under our noses: AI is stepping in to eliminate that uncomfortable “stuck” feeling before kids even get a chance to feel it, and that discomfort is the whole point of learning.
A few weeks back, my own 10-year-old son had a book report on Wonder due. He’d read half the book, dragged his feet on writing, and I turned around to find he’d already asked AI to list three possible themes for the book. He picked “kindness,” copied a few lines, rearranged them, turned it in, and got an A. A week later, we drove past a billboard for the movie adaptation, and I asked him what the main theme of the book was. He stared at me for 10 seconds and said, “I don’t remember, it was something nice.” He’d done the assignment, checked the box, got the grade, but he never did the thinking. The discomfort of sorting through what he actually thought about the book, of sifting through plot points to pull out a theme that mattered to him, AI did that part for him before he even tried.
Cognitive resilience isn’t some fancy buzzword for being tough. It’s the ability to sit with not knowing, to work through confusion, to change your mind when you’re wrong, and to pull a solution out of a messy pile of wrong starts. That muscle only gets built when you go through that discomfort yourself. If you never have to feel stuck, you never learn how to get unstuck. It’s like if you always carry your kid when they get tired on a walk: they never build the endurance to walk a mile on their own.
I don’t think banning AI is the answer, anyway. It’s everywhere, it’s useful, and kids need to learn how to use it responsibly. But I’ve started doing three small, low-effort things with my tutoring students and my own kid that push back on the urge to eliminate discomfort before it starts, none of which require extra hours of work or strict no-phone rules.
First, the 20 minute rule. No matter what assignment they’re working on, before they open any AI tool, they spend 20 minutes writing down every single thing they’ve tried, even the wrong stuff, and every question they have. I tried this with Mia that day. She grumbled, but spent 20 minutes working through that first problem, wrote down two wrong approaches, and figured out the last step on her own. When we pulled up AI’s answer to check, we saw AI had messed up the x-intercept. Mia caught it immediately, because she already knew what it should be. If she’d just taken AI’s answer from the start, she never would have noticed, and she never would have practiced checking someone else’s work against her own reasoning.
Second, ask for messy work first, not just the finished product. For parents helping with homework, or teachers grading assignments, give a few extra points just for the scratch work, the bad brainstorms, the list of wrong ideas. My son’s 5th grade teacher started doing this this year: 10% of the grade for any writing assignment is just turning in your first brainstorm, full of half-baked ideas, before you use any tools. It takes two extra minutes to check, and it forces kids to do the hard thinking part before they outsource it.
Third, when your kid says they’re stuck, don’t tell them to ask AI. Instead, say, “What’s one thing you do know right now?” That’s a tiny phrase that pushes them to sit with the discomfort for one more minute, instead of reaching for a tool right away. I started doing this after the book report incident, and more often than not, my son can name one thing, then another, and before he knows it, he’s working through the problem on his own.
Last week, my son was working on his science fair project, testing how light direction affects bean plant growth. He got confused why all the plants leaned toward the window, and I watched his hand drift toward his phone. Then he paused, pulled out his notebook, and wrote down three ideas he had: more sun, warmer near the glass, the wind from the open window pulls them that way. He still used AI at the end to look up why plants actually lean toward light, but he told me halfway through he was pretty sure it was the sun, and AI proved him right. He was prouder of that guess than any A he’s gotten on a pre-written AI report.
I don’t have all the answers. Some days, I forget the 20 minute rule, some days Mia uses AI to do her homework before our session anyway, and I get it: homework is tiring, kids are busy. The whole cheating debate keeps everyone yelling, but I think the real work is way smaller than that. It’s just saving a little room for the messy, uncomfortable stuck moments that actually teach kids how to think. We’ll figure the rest out as we go.
Please indicate: Thinking In Educating » What Happens to Kids’ Cognitive Resilience When AI Eliminates Productive Discomfort