Is AI Dependency Changing How Students Learn Hard Subjects? A Teacher’s On-the-Ground Look
Last Tuesday, third period pre-algebra, we’re working on factoring quadratic trinomials—everyone’s least favorite unit. I’m walking between scuffed desks, checking in on progress, and stop at Mia’s. She’s been quiet all week, staring at the same problem for 10 minutes, her phone tucked half under her marble-patterned notebook. I lean over, and she doesn’t even flinch or slide the phone out of sight. “I asked AI to solve it,” she says, nodding at the screen. “But I still don’t get how it got that first step.”
I’ve been teaching 8th grade math at a public middle school outside Portland for 12 years. Every parent-teacher conference this year, the same question comes up, laced with worry: Are kids getting too dependent on AI? Is it changing how they approach hard subjects for the worse? I get the fear—any new tool that upends how we’ve always done school feels like a threat. But what I see in my classroom every day doesn’t match the hype.
Ten years ago, if a kid got stuck on the first step of a hard problem, what would they do? They’d skip it. They’d scribble something random to fill the space, copy the answer from the kid sitting next to them, and never mention it again. Most of the time, they couldn’t even name what they didn’t get—they just knew they were stuck, so they quit. They didn’t come after class for help, because admitting you’re that lost felt embarrassing.
Last week, three kids stayed after class for extra help. All three had used AI on their homework. None of them had just copied the answer and called it done. All three brought their AI-generated work, pointed to a specific step, and said “this is where I’m confused.” That’s a shift I never saw 10 years ago, when most kids would rather take a zero than admit they didn’t know where to start.
Take Javi, who’s had an IEP for math processing since 5th grade. He hates word problems more than anything—used to leave the entire word problem section blank on every homework assignment, two years ago. This year, he uses AI to break the long, confusing paragraphs into small, one-sentence questions, then works through each chunk one by one. Last month, he raised his hand to walk through a multi-step word problem in front of the whole class—something he’s never done before. Is he dependent on AI? Sure, in the same way I depend on a calculator for long division when I’m grading 120 quizzes. It’s a tool that takes away the part that stops him from doing the actual thinking.
Of course there are kids who just copy. I have a handful of students who turn in AI-generated homework with no idea what any of the steps mean. But that’s nothing new. When I started teaching, kids copied answers from the back of the textbook, or snapped a photo of a friend’s homework between classes. The percentage of kids who cut corners is roughly the same as it was 12 years ago. We just notice it more now because it’s shiny and new, and it feels like a bigger threat than the old ways of cutting corners.
The question I get most from parents is what they can do at home, to make sure AI isn’t making their kid lazy or dependent. I don’t tell them to ban phones or AI entirely. Banning it just makes kids hide it, so they don’t get the help they need when they’re still stuck. Instead, I tell them to do two simple, low-effort things when their kid is working on a hard subject. First, ask: “What part did AI help you with, and what part are you still working on?” That takes 10 seconds, and it pushes kids to talk through their process instead of just copying an answer and moving on. Second, if they got an answer from AI, ask them to explain the last step back to you in their own words. If they can’t do that, that’s your cue to help them look up a quick video tutorial, or shoot me an email to ask for extra help. It’s not a big interrogation, just a quick check to make sure they’re still doing the thinking.
I did a small trial last school year: first semester I banned AI for homework, second semester I allowed it with only one rule—kids had to write one sentence next to every AI-assisted problem saying what they learned or what they still didn’t get. The number of kids who attempted all hard homework problems went up 28%, and the average quiz score stayed almost exactly the same. The kids who were going to work through the hard stuff still did. The kids who were going to cut corners still cut corners. The difference was that more kids got help when they were stuck, instead of just giving up.
That day after class, Mia and I sat together and walked through her factoring problem. AI had skipped the step where you pull out the greatest common factor first, because the algorithm assumed that was obvious enough to leave out. We wrote it out line by line, and Mia laughed when she realized she’d gotten that part wrong on her own first try, which is why nothing matched AI’s final answer. On last week’s quiz, she got that exact type of problem right.
I still don’t know if what we’re seeing right now will be a net good or bad for student learning 10 years from now. It’s still too new, and we’re all figuring it out as we go. What I do know is that we’re overstating how much AI has changed kids’ ability to work through hard subjects. The hard work of learning has always been showing up when you’re stuck, figuring out what you don’t know, and asking for help. AI just changes where that stuck point falls. It doesn’t remove the work. And for a lot of kids, it makes it easier to start that work in the first place.
Please indicate: Thinking In Educating » Is AI Dependency Changing How Students Learn Hard Subjects