Why I Told My Son to Use a “GPT in His Head” for Homework
The word problem was longer than it needed to be, and Sam, my fifth-grader, was already defeated. He’d read it three times. His pencil hovered. His eyes did that thing where they glaze over and the paper becomes a distant country.
I sat next to him, ready to do what I usually do: ask him to try again, then slowly lose patience, then just tell him the answer. But something stopped me. I’d been using ChatGPT for a few weeks—not for anything important, just recipes, packing lists, explaining why kids get carsick. And I’d noticed something about the way it talked. It didn’t give up. It just asked clarifying questions.
So instead of saying, “What don’t you get?” which is like asking someone to describe a hole they’re standing in, I said, “If you were the chatbot, what would you ask yourself first?”
He looked at me like I’d grown a second head.
“I’m serious,” I said. “You have a little GPT in your brain. Not the part that knows the answer. The part that asks questions. What would it say?”
He blinked. Then, quietly, he said, “What do we already know?”
“Okay. So answer.”
He looked at the problem. “There are four shelves. Each shelf has six books.”
“Good. What else?”
“A book fell off the bottom shelf.”
“What would it ask now?”
He hesitated. “How many books are left?”
“Do you know how to figure that out?”
He did, and he did it. The whole thing took about three minutes, and the only real change was that I stopped being the person with the answer and became the person with the questions. But something bigger stuck with me: he had that inner voice all along. I just had to point it out.
Since that night, I’ve been paying attention to how I think through problems myself. And I’ve realized that I already had a “GPT in my head” long before I ever opened the app. It’s just a set of reflexive questions I ask when I’m stuck.
What am I actually trying to solve here?
What information do I already have?
What’s one small step I can take?
That’s really all it is. Not artificial intelligence, but the habit of prompting your own brain. For a kid, though, that habit doesn’t come automatically. It has to be shown, named, and repeated.
The next time it came up was with his younger sister, Mia, who was writing an apology note to a classmate. She was frozen because she thought an apology had to be perfect. I could see her about to write something stiff and fake, or give up entirely.
Sam was at the table doing his own homework, and I caught myself about to say, “Just say you’re sorry and that you won’t do it again.” Instead, I turned to Mia and said, “Okay, pretend you’re the GPT. What would it ask you?”
She got into it. “What did you do?”
“I called her a name at lunch.”
“What do you want to say to her?”
“I want to say I didn’t mean it, and I feel bad.”
“Then write that.”
She did. It wasn’t perfect, but it was real. Sam looked up and said, “That’s what I did.”
I’m not telling this story because some rubber duck method is a miracle. I’m telling it because I think we forget that our inner voice is a tool that can be shaped. We treat thinking as something that either happens or doesn’t, but it’s actually just a series of prompts, most of which we learned from adults.
So the practical parent part of me started to get more intentional. In the car, when he told me some elaborate story about who said what on the playground, I’d say, “What do you wonder about that?” When he’d say “I don’t know” about why he couldn’t get started on chores, I’d say, “What would the GPT tell you?” Sometimes he’d roll his eyes, but sometimes he’d say, “I’d tell myself to just do the first thing.”
That phrase “first thing” has become our quiet anchor. It doesn’t solve the whole problem. But it gets the pen moving, which is nine-tenths of most battles.
Now, before anyone worries that I’m trying to turn my child into a miniature bot, I’m not. The idea isn’t to make thinking cold or robotic. It’s actually the opposite. A GPT doesn’t have feelings, but it does have structure. When we teach a kid to ask themselves “What do I already know?” we’re giving them a way to access what they actually understand, rather than panicking because they don’t understand everything.
That panic is the real enemy. I see it in Sam. He looks at one complicated problem, or one blank page, and his brain broadcasts a single message: “I can’t do this.” The inner GPT is a way to change that broadcast. Instead of “I can’t,” it says, “Interesting. What information might be useful?”
I’ve started using this on myself too. When I’m stuck writing an email that’s been sitting in my drafts for three days, I’ll literally think, “What would the GPT tell me?” And it’s always something like, “Just write the first sentence and see.” Or when I’m frustrated about something Sam did, I’ll ask, “What is that behavior trying to tell me?” It’s weird how often I already know the answer.
The biggest thing I’ve learned is to speak those questions out loud when the kids are nearby. Not at them, just near enough. They pick them up. A couple of weeks ago, Sam was playing with Legos and couldn’t find a piece. I heard him mumble, “What else do I have that’s the same color?” And I almost did a little fist pump, because that’s a prompt, not a complaint.
He’s not suddenly getting straight A’s. There are still tears over math pages and defensive grunts about essays. But those meltdowns don’t last as long. They used to end with his head down, a stubborn refusal to keep going. Now they often end with him staring at the paper and saying, “Okay, what do I actually need to figure out?”
Sometimes the answer is “I don’t know.” And that’s fine. The GPT also says that. But then it usually follows up with, “What’s one possibility?” So we try that.
I don’t know if this will still work in middle school, or high school, or when he forgets I ever said the word GPT. Maybe it’ll be replaced by a different metaphor, or maybe he’ll just have the questions inside him, no memory of where they came from. That’s the hope, anyway. That eventually he won’t need me to prompt him. He’ll hear a calm voice in his head asking, “What’s the first thing?” and it’ll sound like his own.
Please indicate: Thinking In Educating » Why I Told My Son to Use a “GPT in His Head” for Homework