Is AI Dependency Changing How Students Tackle Hard Subjects? What I’ve Seen As A Tutor
Last Tuesday, I was sitting across from 10th grader Javi at our regular after-school tutoring spot—my kitchen table, half a blueberry Pop-Tart crumbled next to his notebook, his cracked phone with a soccer sticker face-down by my cold coffee. We were knee-deep in quadratic factoring, and Javi had been stuck for seven minutes on the same problem. He huffed, pushed his pencil back, and reached for his phone. “Let me just get AI to show me where I messed up,” he said. “This is taking forever.”
Five years ago, before these tools became ubiquitous, Javi would’ve done one of two things: stared at the page until I offered a hint, or flipped to the back of the textbook to copy the answer and try to reverse-engineer the steps. What he did next wasn’t all that different from what kids have always done when they hit a wall with hard material—look for a shortcut. But the ease and speed of it made me pause, like it does every time I see it. Everyone’s talking about AI dependency these days, warning that kids can’t do anything on their own anymore. But after three years of tutoring middle and high school kids through everything from algebra to AP bio essays, I’m not sure we’re talking about the right thing.
Last month, I was visiting my sister, and my 20-year-old niece Lila, who’s studying biochemistry, was working on a lab report at their dining table. I watched her pull up AI to draft the methods section, the boring part where you just list out all the supplies and steps you followed in the experiment. She tweaked it in five minutes, then turned her full attention to the results section, where she had to analyze what her data actually meant. “If I didn’t have AI to do the repetitive stuff,” she told me, “I’d be up until 2 a.m. instead of 11. I don’t use it to do the hard thinking, just the busy work.”
That contrast sticks with me. What I’ve noticed is that AI didn’t create a new problem—it amplified an old one. The kids who would’ve avoided struggling through a hard problem by any means necessary 10 years ago now just use AI instead of copying from a friend or faking their way through a lab report. The kids who want to actually learn the material use AI as a tool, not a crutch. But that doesn’t mean nothing changed. The shift in how they approach difficulty is real, it’s just not the “kids are all lazy now” narrative you see online.
Before AI, if you got stuck, you had to put in work to get an answer. You had to flip through notes, re-read the chapter, email your teacher, wait until the next day to ask. That 10 or 15 minutes of struggling? That’s where you make connections in your brain, that’s where you learn what doesn’t work so you remember what does. Now, you can get a full step-by-step in 30 seconds. So kids don’t build that habit of sitting with discomfort anymore. That’s the real change, not some vague “dependency”—they’re out of practice with struggling on their own first.
After Javi pulled up the AI walkthrough that Tuesday, he nodded along, said “oh right, that’s where I messed up,” and went to close the app to move on. I stopped him. I’ve had this rule for two years now, after I noticed this same pattern with half my tutees: before you look anything up on AI, you have to write down at least two different attempts you made to solve the problem, and one sentence about why you think each didn’t work. Then, after you get the AI answer, you have to explain the tricky step back to me in your own words, no reading off the screen.
That Tuesday, Javi tried to explain the factoring step, and got tripped up halfway. He’d thought he understood what AI was saying, but he’d missed why we factor out the negative sign in the first place. That’s the mistake he would’ve made on the test, too, if we’d just moved on. We worked through it together for another five minutes, and that’s the mistake he hasn’t made again on a quiz.
What I tell parents who ask me about this is that banning AI at home doesn’t work. It’s everywhere, kids will just use it when you’re not looking. Instead, try two small, concrete things: First, agree on a “two attempts” rule before any AI use. Your kid doesn’t have to solve the whole problem, just write down what they tried. Second, after they use AI, ask them one simple question: “Can you show me the part you got stuck on, and explain how the answer fixed it?” That’s it. No big lecture, no checking every single problem, just those two steps. It takes 30 extra seconds, but it forces them to do the hard thinking instead of just copying the result.
So are we overstating AI’s impact on how kids approach hard subjects? I think in some ways, yes. We’re acting like this is a complete upheaval of education, when really it’s just a new tool that changes how shortcuts work. The kids who cared about learning still care, the kids who didn’t still don’t. But it’s not nothing. The loss of practice sitting with hard, uncomfortable stuckness is a real shift. Kids don’t get that 10 minutes of trial and error anymore, because they don’t have to. That changes how they approach hard subjects long term—when they hit a problem on a test where they can’t use AI, they freeze faster than kids did 10 years ago, because they haven’t practiced getting unstuck on their own.
The two-attempt rule works for most of my tutees, Javi included. He’s gotten better at working through the first two tries on his own before reaching for his phone. But there are still days when he forgets, when he’s tired after soccer practice and just wants to get his homework done, and I have to remind him to pull out his notebook and write down what he tried. We’re all adjusting, slowly. It’s not a crisis, it’s not nothing. It’s just a small shift that needs small adjustments, nothing more. Some days that feels like enough, and some days I wonder what it’ll look like for the next group of kids coming up. There’s still no clear answer, but we’re figuring it out as we go.
Please indicate: Thinking In Educating » Is AI Dependency Changing How Students Tackle Hard Subjects