On a humid morning in Ubud, a ten year old named Lila sat cross legged on the bamboo floor of our outdoor classroom and stared at a gecko.
The assignment was simple. Choose something alive on campus and make a field guide page. Description, behavior, habitat, and one question you still had.
Lila already had her creature. The gecko lived behind the whiteboard and darted out whenever a fly got too close.
She opened her laptop, typed a single sentence into an AI tool, and watched a polished entry appear.
Scientific name. Diet. Distribution. Predators. Fun facts.
It looked like a field guide page. It sounded like a field guide page.
It was also unrelated to the gecko in front of her.
Her teacher, Ms. Putu, walked over and asked the question that reveals everything in the AI age.
“What did you notice?”
Lila blinked.
She could read the entry back. She could not tell you what she had actually seen.
The problem was not that she used a tool. The problem was that the tool let her skip the best part of the assignment, the part that builds a mind. Looking closely enough to wonder.
In a world of instant answers, the scarce skill is not knowing. The scarce skill is staying curious long enough to learn.
Protect the question.
We used to live in a world where information was harder to get. Curiosity was rewarded because it was the engine that pushed you toward books, people, experiments, and places.
Now answers are everywhere. They are fast. They sound confident. They often sound smarter than the person asking.
When answers are cheap, something else becomes expensive.
Attention.
Integrity.
Questions that actually lead somewhere.
Curiosity is not a personality trait. It is a practice. It is the ability to keep a question open even when you could close it with a quick, clean response.
Children are not losing curiosity because they are lazy. They are losing it because the environment keeps rewarding closure.
A child asks a question. An adult answers quickly. A screen answers faster. An AI answers in a way that feels complete.
I watch shoulders drop when the answer arrives. The social pressure drops too. The discomfort of not knowing disappears.
The brain learns a lesson. The goal is relief, not understanding.
There is research that helps explain why this is such a strong pull. In a classic study, people who expected to be able to look up information later remembered less of the information itself and more of where to find it.[1]
That is not a character flaw. It is efficiency.
But childhood development is not always about efficiency. It is about building inner capacity. Attention. Memory. Curiosity. Agency.
Newer work suggests a similar risk with AI writing assistance. A preprint comparing people who wrote with an AI assistant to those who wrote without it found differences in brain activity and in how much people felt ownership of what they produced.[2]
That should make every parent and educator pause.
The danger is not that AI helps. The danger is that it helps too early.
When we give a child a tool before they have made contact with the thing itself, we are not just making life easier. We are training a habit of skipping wonder.
Curiosity feels alive.
Consumption feels full.
Curiosity says, “What is really going on here?”
Consumption says, “Give me something that sounds done.”
The two can look similar from the outside. Both might involve a screen. Both might involve research. Both might involve information.
The difference is what happens next.
A curious child ends with more questions and a clearer sense of what they do not yet know.
A consuming child ends with a paragraph they could paste into a document and a feeling of being finished.
In the AI age, we are going to have to teach children the skill of not being finished too early.
Part of the challenge is that school can accidentally train answer mode. In her work on curiosity in schools, Susan Engel describes how curiosity can be nurtured or suppressed by the learning environment, and how classrooms can drift toward right answers over student inquiry.[3]
The good news is that curiosity is trainable. A recent study suggests that practicing question asking can foster aspects of curiosity in young children learning science, especially for kids with less background knowledge.[4]
A small story about a big upgrade
Later that week, we tried the same assignment again, but with one change.
Before anyone touched a device, Ms. Putu walked the class outside and gave them five minutes of silence.
“Choose your living thing,” she said, “and write down ten observations. No facts. Only what you can see, hear, smell, or feel.”
The campus became a slow motion film. Children crouched near puddles and watched ants build bridges. They traced leaf veins with their fingers.
When they came back, the room felt different. Not quieter. More awake.
Lila chose the same gecko. This time she wrote:
“It freezes when people look at it.”
“It moves like a small machine.”
“It stays near light at night.”
“It has a loud throat when it calls.”
Then Ms. Putu gave them the next instruction.
“Now you can use a tool,” she said. “But you are not allowed to ask for an entry. You have to ask for help with a question.”
Lila typed, “Why do geckos stay near lights at night?”
The answer that came back did not end her work. It opened it.
She learned about insects and heat and hunting patterns. She also formed a hypothesis. Then she tested it. She counted geckos near bright lights versus dark corners.
That is the sequence we want.
Observation first.
Questions second.
Tools third.
And real world verification whenever possible.
Curiosity is the steering wheel. AI is the engine.
If you want one simple filter for healthy tool use, use this.
Call it the next step test.
If the tool ends the work, it is probably replacing development.
If the tool creates the next step, it is probably amplifying learning.
The Question Upgrade
Here is a tool you can use at home or in a classroom. It takes a child’s first question, usually the shortest version, and upgrades it into a question that can lead to learning rather than just an answer.
The goal is not to make children sound sophisticated. The goal is to keep their minds in the driver’s seat.
Start with whatever question your child naturally asks.
“What is photosynthesis?”
“Who was Julius Caesar?”
“Why is my friend being mean?”
Then move it up the ladder using these five upgrades.
Upgrade one. Make it specific.
What part are you stuck on? What do you need to know next?
Upgrade two. Make it causal.
Why does it work that way? What causes what?
Upgrade three. Make it comparative.
Compared to what? What changes if we change the context?
Upgrade four. Make it testable.
How would we find out? What evidence would count?
Upgrade five. Make it personal.
Why do you care? What does this connect to in your life?
You do not need all five every time. One upgrade can reopen thinking. Two is often enough.
Here is what it looks like with Lila’s gecko.
Start. Tell me about geckos.
Specific. Why do geckos stay near the lights on our campus at night?
Causal. Is it because insects gather near light, or is something else going on?
Comparative. Are there more geckos near bright lights than near darker corners?
Testable. How could we count and compare in the same place, on the same night?
Personal. What could I notice tonight when I walk past the lights?
The upgrade creates a path. It turns a request for information into an invitation to investigate.
Two question builders you can borrow
If you like structure, there are two classroom routines that translate beautifully to home life.
The Question Formulation Technique is a simple protocol for generating, improving, and prioritizing questions.[7] It gives kids a way to move from “tell me the answer” to “help me investigate.”
Harvard’s Project Zero uses short thinking routines like See, Think, Wonder. It is a three step move. What do you see? What do you think is going on? What does it make you wonder?[8]
Both routines do the same thing Ms. Putu did for Lila.
They slow the moment down.
They keep observation in front.
They make questions the work.
How to use AI without outsourcing curiosity
Most parents do not actually mind their child using AI.
What they mind is the feeling that their child is disappearing behind it.
The fix is not surveillance. It is structure.
Here are four ways to make AI serve curiosity instead of replacing it.
One. Use AI as a tutor, not a ghostwriter.
Ask it to teach a concept step by step, then quiz your child, then correct misunderstandings.
Two. Use AI as a question generator, not an answer dispenser.
Ask it for ten questions about a topic, then have your child choose the one that feels most interesting or most important.
Three. Use AI as a debate partner.
Ask it to challenge an argument, offer counterexamples, or steelman the other side.
Four. Use AI as a planning assistant for real world action.
Ask it to propose an experiment, a prototype, an interview plan, or a field observation checklist.
Then have your child do the human part. Notice, decide, test, and revise.
There is a deeper reason this matters. Neuroscience research suggests curiosity changes how the brain learns. When people are genuinely curious, they tend to remember the answer better, and they often remember other information they encounter while in that curious state.[5]
That is a quiet reason to protect the wonder gap before an answer arrives. If we fill every gap instantly, we remove one of the brain’s best learning modes.
A prompt pack for inquiry
If you want a child to use AI well, teach them to ask for the kind of help that keeps them thinking.
Here are prompts you can keep on a note in your kitchen or classroom. They turn AI into a coach instead of a replacement.
“Before you answer, ask me five clarifying questions.”
“Give me three possible explanations, and tell me what evidence would support each one.”
“Teach this in a simple way, then give me a harder version, and ask me what I understood.”
“What is a common misconception about this topic?”
“What would someone intelligent disagree with here?”
“Suggest a simple experiment I can do with household materials.”
“If we changed one variable, what would likely change, and why?”
“Help me write an outline, but do not write any sentences. Ask me what I want to say in each section.”
When children learn these prompts, they stop using AI like a vending machine.
They start using it like a workshop tool.
Signs the tool is doing too much
You do not need to become a detective. You just need to know what to look for.
When AI is stealing curiosity, you will often see one of these signs.
Your child cannot explain what they just made.
They use phrases they never use out loud.
They finish quickly but feel oddly empty or irritable.
They avoid follow up questions, because follow up questions would reveal the gap.
They treat learning like a transaction. Get output, get reward, move on.
In the essay writing study cited earlier, participants who relied more on the AI assistant reported lower ownership and struggled more to accurately quote what they had submitted.[2]
That maps neatly onto what parents notice. Polished output that the child cannot explain.
The fix is not shame. It is restoring the sequence. Attempt first. Tools second. Explanation always.
If you see it, do not shame.
Shame teaches secrecy.
Instead, return to the basics.
Observation.
A better question.
A small next step in the real world.
Boredom is part of the curriculum
If you want to know where curiosity lives, look for the moment right before entertainment arrives.
That moment is usually boredom.
Boredom is not a symptom of a broken child. It is often the beginning of a question.
When children have a few minutes of empty space, their minds do what minds evolved to do.
They scan. They notice. They imagine. They invent a problem to solve.
The modern environment does not allow that process to finish.
A phone can fill the gap in two seconds. An AI can fill it with something that looks productive.
Studies suggest that a boring task can increase creative idea generation afterward. One likely mechanism is simple. Boredom nudges the mind into daydreaming and mental wandering, which often produces fresh connections.[6]
So when a child says, “I’m bored,” they are not reporting a defect. They are standing at the doorway of invention.
At our school we sometimes call boredom “the doorway.” We do not worship it. We simply stop rescuing children from it too fast.
A practical family rule is a short pause before screens or answers.
Three minutes.
When your child says, “I’m bored,” you can say, “Good. That means your brain is about to make something. Give it three minutes.”
If they cannot find a next step, you can offer a prompt that points them back to the real world.
“What is one thing you could notice?”
“What is one thing you could build?”
“What is one thing you could ask?”
Curiosity is not always a spark. Sometimes it is a slow burn that needs a little oxygen.
A question wall, not a performance wall
In many schools, the walls are covered with finished work. Perfect handwriting, polished posters, neat conclusions.
That can be beautiful. It can also send a quiet message.
We celebrate what is complete.
In the AI age, we also need places that celebrate what is incomplete.
At Empathy School we keep a Wonder Wall. It is a board where students post questions they cannot answer yet.
The questions are messy.
“Why do people lie?”
“How do turtles know where to go?”
“Why do I get embarrassed so fast?”
“What makes something fair?”
Once a week a class chooses one question and treats it like a project. Sometimes they use books. Sometimes they interview someone. Sometimes they run an experiment. Sometimes they use AI, but only after the question is already alive.
The wall does something subtle and powerful.
It makes not knowing normal.
When not knowing is normal, children do not need to hide behind polish.
They can be beginners in public.
That is where real learning starts.
A teenager’s shortcut, and the quiet cost
One of our older students, Rafi, was assigned a short presentation on an environmental issue he cared about.
He chose plastic in the ocean. He cared, in theory. In practice, he was exhausted and behind.
He asked an AI tool for a presentation. He copied the structure, the facts, the confident tone.
He delivered it smoothly.
Then a classmate asked a question.
“Where does the plastic here actually come from?”
Rafi froze.
He was not lazy. He was not dishonest in the usual sense. He was simply disconnected.
So we did the only thing that fixes that problem.
We sent him into the world.
He came back quieter.
He talked to a market vendor, a cafe owner, and a woman who manages waste in her neighborhood.
His next presentation was less impressive on paper. It was also real.
It included a map he drew of where trash collects after rain.
It included one uncomfortable fact he could not unsee. Sometimes the plastic problem is not ignorance. It is infrastructure.
He ended with a question he could not answer.
“What would make it easier for people to choose differently?”
That is the kind of question that leads to agency.
AI can write a presentation.
It cannot create a relationship between a teenager and the real world.
The moment kids reach for answers
When kids ask for instant answers, it is usually not because they love shortcuts.
It is because they are trying to escape one of three uncomfortable feelings.
Confusion.
Boredom.
Fear of being wrong.
If you respond only with rules, you will miss the real need.
The better move is to treat “I want the answer” as a signal.
It is your child saying, “I do not know how to stay with this.”
That is not a moral failure. It is a skill gap.
So you build the skill.
You slow down.
You help them ask a better question.
You help them tolerate being unfinished.
Then you let the tool help, without taking the steering wheel.
A script for the kitchen table
When your child says, “Can I just ask AI?” you can try this sequence.
First. “Tell me what the assignment is asking in your own words.”
Second. “What do you already know?”
Third. “What is one thing you are genuinely curious about?”
Fourth. “Let’s write three better questions.”
Only then. “Okay. Now we can use a tool, but we will use it to help with your question, not to replace your thinking.”
This takes two minutes. It turns AI use from a reflex into a practice.
Age matters
The same principle applies at every age. Protect the question, then use tools to deepen it.
But the practices should match the child.
For younger children, curiosity is physical.
They ask with their hands. They learn by touching, building, spilling, and looking again.
Your job is to say, “Let’s find out,” more often than you say, “Here’s the answer.”
For middle childhood, curiosity becomes social.
They want to know what others think. They want to test ideas. They can hold a question open for longer.
Your job is to help them build routines. Question lists, small research plans, simple experiments, and reflection.
For teenagers, curiosity becomes identity.
They are asking, “Who am I?” “What do I believe?” “Where do I belong?”
Your job is to treat those questions as worthy of real conversation, not something to outsource to an algorithm.
A teenager can use AI as a mirror, but it cannot be the only mirror.
Real identity is formed in real relationship, with real consequences and real repair.
The verification habit
One more skill becomes essential when answers come easily.
Verification.
Children should learn early that confident answers are not always true.
This is not about distrust. It is about maturity.
We teach children to look both ways before crossing the road. Not because cars are evil, but because roads are real.
Information is now like that.
OpenAI notes that ChatGPT can produce confident but incorrect information.[9]
Research surveys on hallucinations in large language models echo the same point. The tool can be persuasive and wrong.[10]
So “How would we check that?” is the most future proof question you can teach.
A simple family practice is to ask one question whenever AI provides a fact.
“How would we check that?”
Sometimes the answer is a book. Sometimes it is a different source. Sometimes it is an experiment. Sometimes it is asking a real person who has lived it.
The point is not to turn children into skeptics.
The point is to turn them into adults who do not confuse fluency with truth.
FAMILY EXPERIMENT
The Question Upgrade Ritual
The goal is to make better questions a weekly habit, so your child does not outsource curiosity when schoolwork or life feels hard.
Set aside ten to fifteen minutes once a week. Choose a calm time. Phones away.
Step one
Pick one topic your child is currently touching. A book, a science unit, a news story, a hobby, a friendship problem, a project idea.
Step two
Ask for the first question that comes to mind. Write it down exactly as they say it.
Step three
Upgrade it using the five upgrades. Specific, causal, comparative, testable, personal.
You do not need all five every time. Two is often enough.
Step four
If you want to use AI, use it in one of these roles.
Tutor. Teach me this, then quiz me.
Question coach. Give me ten better questions about this.
Debate partner. What would a smart person disagree with here.
Planner. What is a simple experiment or observation plan I could do this week.
Step five
End with two sentences from your child.
“One thing I learned is…”
“One question I still have is…”
If your child can end with a living question, the ritual worked.
A script for parents
“In this family, we use tools. But we do not use tools to skip becoming capable. Your job is to bring the question. The tool can help with the search.”
What to notice this week
One moment when your child reached for an instant answer.
One moment when you helped them stay with the question instead.
A school question
“Where do students practice asking questions that cannot be answered with a single search, and how do you teach them to use AI as an inquiry tool rather than a shortcut?”
Closing
Lila’s second field guide page was not as polished as the first.
It also belonged to her.
It included smudged pencil marks from sitting in the grass. It included a diagram of where the gecko hid. It included a question at the bottom.
“Do geckos recognize people?”
That question mattered more than the facts.
Facts can be generated.
A living question is a sign of a living mind.
Protect the question.
Endnotes
[1] Sparrow, B., Liu, J., and Wegner, D. M. Google Effects on Memory: Cognitive Consequences of Having Information at Our Fingertips. 2011. Science. Volume 333, issue 6043, pages 776 to 778.
[2] Kosmyna, N., Hauptmann, E., Yuan, Y. T., Situ, J., Liao, X. H., Beresnitzky, A. V., Braunstein, I., and Maes, P. Your Brain on ChatGPT: Accumulation of Cognitive Debt when Using an AI Assistant for Essay Writing Task. 2025. arXiv preprint 2506.08872.
[3] Engel, S. Children’s Need to Know: Curiosity in Schools. 2011. Harvard Educational Review. Volume 81, issue 4, pages 625 to 645.
[4] Park, A. T., Colantonio, J., and colleagues. Question asking practice fosters aspects of curiosity in science content in young children. 2025. npj Science of Learning. Locator: article 00384.
[5] Gruber, M. J., Gelman, B. D., and Ranganath, C. States of Curiosity Modulate Hippocampus Dependent Learning via the Dopaminergic Circuit. 2014. Neuron. Volume 84, issue 2, pages 486 to 496.
[6] Mann, S., and Cadman, R. Does Being Bored Make Us More Creative. 2014. Creativity Research Journal. Volume 26, issue 2, pages 165 to 173.
[7] Right Question Institute. What is the Question Formulation Technique. No date. Locator: web resource describing the four rules and the improve and prioritize steps.
[8] Project Zero, Harvard Graduate School of Education. See, Think, Wonder. No date. Locator: web resource describing the routine.
[9] OpenAI Help Center. Does ChatGPT tell the truth. No date. Locator: help center article on accuracy and limitations.
[10] Huang, L., and colleagues. A Survey on Hallucination in Large Language Models. 2023. arXiv preprint 2311.05232.