Productive Struggle Matters Even More in the Age of AI

Cody Carlton
Jul 14
4 min read

Every teacher knows the look.

I taught 9th grade ELA for several years, and there's one class I still think about. We were working through The Book Thief, specifically the way Death describes a color every time someone dies. One student, I'll call her Mary, kept getting stuck. She'd try an idea, cross it out, raise her hand, put her head down, and try again. This went on for a good ten minutes.

Then she looked up and said, "Wait. He's not just describing the sky. He's telling us how he copes with all of it." I could see it on her face before she even finished the sentence.

That moment wasn't just relief. It was ownership. Mary didn't just get the answer. She made sense of it herself.

That's the part of learning we can't afford to lose right now.

AI can do a lot of good things for students. It can explain, summarize, organize, brainstorm, translate, adjust reading levels, and give kids access to support they might not have had before.

Used well, it makes learning more accessible and a lot less intimidating.

But here's what needs to be said out loud: AI can also make it too easy to skip the thinking altogether.

The Problem with Instant Answers

A correct answer isn't always proof that learning happened. That's always been true, but AI makes it a lot harder to notice.

A student can hand in a polished paragraph, a solved problem, a clean summary, a reflection that sounds thoughtful, all in seconds. On the surface, it looks done.

But can they explain it? Can they do it again without the tool? Can they take the idea and use it somewhere new? Can they tell you where they got stuck and how their thinking changed along the way?

That's where you actually see the learning.

Students are going to use AI. The real question is whether the tool is helping them think, or helping them avoid it.

Productive Struggle Isn't "Just Let Them Figure It Out"

People sometimes hear "productive struggle" and picture kids sitting in confusion until inspiration strikes. That's not it, and honestly, that version doesn't help anyone.

Productive struggle has support built in. Students get a task that's actually worth thinking through, plus enough of a starting place to try something. Mistakes get treated as part of the process instead of proof they're bad at this. And the educator stays close, watching, asking questions, offering a hint, stepping in once the struggle stops being useful and starts being just frustrating.

There's a real difference between a student thinking "this is hard, but I can try something" and a student thinking "I have no idea what to do, and no one is helping me." One builds confidence. The other just wears kids down.

If you want to think more about this, Edutopia's "If You're Not Failing, You're Not Learning" is worth the few minutes it takes to read. Good reminder that failure only teaches anything when it's designed with care and followed by real guidance, not just left alone.

Where AI Actually Fits

Here's one simple shift: AI doesn't have to be the first stop.

Before students open the tool, have them make an attempt first. It doesn't need to be good. It doesn't even need to be close. They just need to put something down: on the page, in the margin, in a discussion, in their own words, before AI enters the picture.

That first attempt gives their brain something to work with. It also gives you something to respond to.

Instead of asking AI to just solve the problem, a student can ask for one hint without the answer attached to it. Or skip the "write this for me" and ask what would make the claim stronger. A summary request can become "quiz me on this after I read it" instead.

Small changes. But they turn AI into something closer to a coach than a shortcut.

OpenAI's Study Mode is a decent example of this shift in practice. It's built to walk students through a problem step by step instead of just handing over the answer. It still needs guardrails and guidance around when students should use it, but the direction is right: help students learn something, not just finish something.

A Small Shift to Try

One way to start: a quick "think first, AI second" routine. Before opening the tool, students answer three questions.

What do I already know?
Where am I stuck?
What kind of help do I actually need?

That last one matters more than it looks. Help isn't one thing. Sometimes a student needs a hint. Sometimes it's a vocabulary word they don't know. Sometimes they need someone to push back on their thinking, or just to try once more before any help shows up at all.

Naming that difference helps students get a little more aware of their own process. It also slows down the reflex to let the tool take over too fast.

If your team is still working out shared expectations around AI, the TeachAI AI Guidance for Schools Toolkit is a solid place to start. Even outside a traditional school setting, the questions in it can help educators think through student agency, academic integrity, and what kind of AI support actually supports the learning goal.

The Aha Moment Still Matters

That "aha" moment isn't just a nice classroom moment. It's evidence something actually shifted: a connection made, a pattern noticed, an answer that finally makes sense instead of just sitting on the page.

I think about Mary a lot these days. That moment with the colors didn't come from a hint sheet or a fast answer. It came from her sitting with it long enough to find it on her own.

AI can support that moment. It shouldn't steal it. If a student gets the answer before wrestling with the idea at all, they might finish the assignment without ever really learning from it.

That's the part worth protecting. Not by keeping every new tool out of the room, and not by pretending students won't use AI anyway, but by designing learning where they still have to think, try, get it wrong, revise, and explain themselves.

Because a completed task isn't really the goal. The goal is a student who can say "I get it now" and actually mean it.