A ten year old named Luca was supposed to bring the measuring tape to the garden.
It was not a dramatic responsibility. It was one tool in a bucket of tools. But the measuring tape was the difference between a plan and a guess.
When the class reached the garden, Luca patted his pockets, looked at his hands, and went still.
“I forgot.”
His eyes moved toward his teacher the way kids’ eyes often move when they want a problem to disappear.
At most schools, an adult would solve it fast. A spare tape would appear. The lesson would continue. Luca would be relieved. The group would move on.
At our campus, the teacher did something quieter.
“Okay. What do you want to do now?”
Luca looked confused. The question was not, “How could you forget” or “Why are you irresponsible.”
The question was, “Are you in the driver’s seat of your own life, even when you make a mistake.”
He thought for a long moment.
“I can run back and get it. Or I can ask if someone has one. Or we can do another job until I come back.”
The teacher nodded.
“Good. Choose one.”
Luca ran.
No one praised him for forgetting. No one shamed him for forgetting.
He simply owned the next move.
That is agency.
Agency is not independence
Parents often hear “agency” and think it means leaving a child alone to figure it out.
That is not agency. That is abandonment with a nicer word.
Agency is a child who knows what they are responsible for, and can take the next step without needing to be pushed, bribed, or rescued.
A child with agency can ask for help without dumping the whole problem on someone else.
They can admit a mistake without collapsing into shame.
They can make a plan and follow it, even when they do not feel like it.
In the age of AI, this capacity matters more than ever, because the modern world is getting very good at removing the need for ownership.
Quick nerdy sidebar (optional, but useful): psychologists have been studying this exact skill for decades, even if they don’t always call it “agency.” One of the most useful frameworks is Self Determination Theory, which says kids tend to thrive when three needs are supported: autonomy (I get a real say), competence (I can handle hard things), and relatedness (I belong). [1]
Notice how that matches the chapter’s whole point: agency is not “do whatever you want.” It’s “you get a say, and you also own the consequences.” Studies on autonomy supportive parenting suggest that when parents offer meaningful choice inside structure (instead of controlling or guilt driven pressure), kids show better day to day well being. [2] And when parenting slides toward psychological control by manipulating kids with shame, guilt, or love withdrawal. Outcomes tend to get worse. [3]
The world is training children to be passengers
A passenger life feels smooth.
Adults manage the calendar. Algorithms manage the feed. Notifications manage attention. GPS manages navigation. Auto fill manages writing.
Now AI offers to manage something deeper: the uncomfortable work of choosing.
A child can ask a bot what to eat, what to say, what to write, what to believe, and how to respond when they feel hurt.
Used well, this can be support.
Used early and often, it becomes a quiet theft.
Because the moment a tool becomes the decider, the child practices being a consumer of choices instead of a maker of them.
Children become fluent in outputs while staying clumsy with ownership.
They can look competent while feeling strangely helpless.
The agency gap shows up in small sentences
Listen for these phrases.
“I don’t know.”
“Tell me what to do.”
“Can you just do it for me?”
“It’s not my fault.”
Those sentences are not moral failures. They are data.
They often appear when a child has been rescued so many times that they no longer expect themselves to steer.
Sometimes they appear because the child is anxious and wants certainty.
Sometimes they appear because the child has learned that if they wait long enough, an adult will carry the burden.
AI can deepen this pattern, because it is the perfect rescuer.
It is always available. It never sighs. It never gets impatient. It offers a plan instantly.
If a child uses it as a shortcut for self leadership, the muscle of agency does not get built.
The rescue reflex
Most parents do not mean to weaken agency. We do it because we are tired, we care, and we want the house to run.
A child forgets their sports shoes. You drive back home to get them.
A child forgets their homework. You message the teacher.
A child procrastinates on a project. At ten at night you sit down and start printing, cutting, and gluing.
In the moment, this feels like love.
In the long run, it can teach a child a dangerous lesson.
If I stall, someone else will carry it.
We also rescue because we are afraid of consequences.
We do not want our child to feel disappointment.
We do not want teachers to judge them.
We do not want them to be “behind.”
But agency is built through consequence, not through lectures.
Consequence does not have to be harsh. It just has to be real.
Consequences are information, not punishment
A consequence is what happens when a choice meets reality.
Punishment is what happens when an adult adds suffering to force obedience.
Agency grows when children meet reality and learn they can handle it.
This is why natural consequences, when they are safe, are powerful teachers.
Another small research backed truth: when adults rescue constantly, kids don’t just lose skills. They lose belief. A systematic review of “helicopter parenting” found that many studies link more overinvolvement/overprotection with higher anxiety and depression symptoms (with an important caveat: most studies are correlational, so we should be humble about cause and effect). [4]
Overprotection also shows up in the research as a predictor of certain “early maladaptive schemas” basically, the brain’s sticky stories about the world like “I’m not capable,” “The world is dangerous,” or “I need someone else to handle things.” Those stories are not destiny, but they’re a helpful warning sign. [5]
And yes, there’s also evidence that a life with zero struggle isn’t the goal. One large study found that people with some adversity (not none, not overwhelming) often reported better resilience and well being than people with either no adversity or very high adversity. The point is not “make life hard.” The point is: don’t erase every manageable consequence. Let your child build the muscle while you stay close. [6]
If a child forgets their water bottle, they feel thirsty for a while and learn to pack it next time.
If a child wastes their allowance, they cannot buy the thing they want later.
If a teenager uses AI to write something they cannot explain, they feel the discomfort of being exposed by a simple question.
You can support your child through that discomfort without removing it.
Support sounds like this.
“That’s hard. I’m here. And I’m not going to erase it for you.”
Three kinds of help
When your child struggles, you have three options.
Rescue: you solve it for them.
Control: you demand compliance until they do it.
Coach: you stay connected while keeping the ownership with them.
Rescue feels kind but can create passivity.
Control can create obedience, but obedience is not self leadership.
Coaching is slower. It requires you to tolerate your child’s frustration without making it your emergency.
It sounds like questions.
“What’s the next step?”
“What are your options?”
“What do you want to choose?”
“What might happen if you don’t do it?”
Coaching also includes clear boundaries.
You are not negotiating whether responsibility exists. You are helping your child choose how to meet it.
AI can be an agency amplifier or an agency substitute
I want to be clear: I am not against tools.
A child can use AI to explore ideas, practice a language, generate study questions, or get feedback on writing.
That can strengthen agency, because the child is still steering.
One more sciencey lens that helps in the AI era: “cognitive offloading.” Humans have always offloaded thinking. We use shopping lists, calendars, and sticky notes because brains are allowed to be human. A major review describes how we offload when a task feels heavy, and how that changes what we store internally versus externally. [7]
The classic “Google effect” study found that when people expect information to be easily available later, they tend to remember the path to the info more than the info itself. [8] That’s not a disaster. It’s a trade off. But it becomes a parenting issue when the offloaded thing is exactly the muscle your child is supposed to be building (starting, planning, revising, apologizing, deciding).
We see a similar trade off with navigation: studies have linked heavier GPS reliance with weaker spatial memory/wayfinding when people have to navigate without it. [9] AI is basically GPS for thinking. Helpful. And potentially skill shrinking if it becomes the default.
Also, this isn’t theoretical anymore. Pew Research Center reports that about a quarter of U.S. teens say they’ve used ChatGPT for schoolwork, and the share increased compared to 2023. [10] So yes: your child is living in the environment this chapter is describing.
But there is a line where AI stops being a tool and becomes a parent.
You can see the line with one simple test.
Is your child using AI to expand their choices, or to avoid choosing at all?
Here are two common ways AI becomes a substitute for agency.
First: the tool becomes the initiator.
The child waits until the bot tells them what to do, then follows the script.
Second: the tool becomes the shield.
The child uses AI to avoid discomfort: difficult homework, awkward conversations, apologies, honesty.
If a child never practices initiating and repairing, they will grow up with lots of words and little spine.
A story about a perfectly written apology
A thirteen year old named Mina once hurt a friend’s feelings during a group project.
It was not cruelty. It was the ordinary sharpness that happens when kids are tired, competitive, and unsure how to say what they need.
Later, Mina wanted the problem to go away.
She opened an AI tool and typed something like, “Write an apology to my friend for being rude.”
The bot produced a flawless message. Warm. Mature. Emotionally intelligent.
Mina copied it and sent it.
Her friend replied, “Thanks,” and the conversation ended.
From the outside, it looked like repair.
From the inside, Mina felt nothing change.
Why?
Because she had not done the work of repair. She had outsourced it.
Real repair requires three things that cannot be autocompleted.
Ownership: I did this.
Impact: It affected you like this.
Change: Next time, I will do this instead.
AI can help a child find words after they do that inner work.
But if the bot does the inner work for them, the child stays the same person, just with nicer sentences.
A better way to use the tool
If Mina had come to an adult first, a teacher might have said:
“Write what you mean in your own words first. Then we can make it clearer.”
The rule is simple: the child leads, the tool supports.
For writing, that might mean: no AI for the first draft.
For relationships, that might mean: no AI for the first feeling.
Your child can ask a bot for wording help after they can say, out loud, what they want to own.
A useful prompt for kids is this:
“Help me say this clearly without changing what I mean.”
That prompt keeps the agency with the child.
Give fewer choices, but make them real
Many parents swing between two extremes.
They either control everything, or they offer endless options and call it freedom.
Neither builds agency.
Agency grows when children live inside a structure they can actually manage.
A good rule of thumb is: offer two or three choices, all of which you can live with.
Then let the child choose.
Here are small examples that build the muscle.
You can do homework before snack or after snack. You choose. Either way, it is done by 7:30.
You can take a shower now or after dinner. You choose. Either way, you go to bed clean.
You can use AI to brainstorm ideas or to check your draft. You choose. Either way, you can explain what you wrote.
Notice what makes these choices different.
They give autonomy inside responsibility.
The child gets to steer the how, while the parent holds the why and the boundary.
Agency requires calm follow through
A choice without follow through is not a real choice. It is a suggestion.
Children learn agency when the world responds consistently to what they do.
That does not mean you need to be harsh.
It means you need to be steady.
When you set a boundary and then collapse under whining, you teach a child that persistence is a way to control you.
When you set a boundary and enforce it with anger, you teach a child that power is scary and relationships are unsafe.
The sweet spot is calm consistency.
Calm consistency is a form of respect.
It says: I believe you can handle reality.
It says: I do not need to win. I need to teach.
How we practice agency at school
At Empathy School, we build agency in unglamorous ways.
Children have jobs that affect other people.
Someone feeds the rabbits. Someone checks the tool shed. Someone sets up the circle. Someone manages the compost.
Why the “jobs that matter” piece works: there’s longitudinal research suggesting that kids who regularly did household chores in early elementary school later showed higher self competence, prosocial behavior, and self efficacy. [11] Translation: responsibility doesn’t just make your house run. It quietly teaches a child, “I’m capable, and I’m needed.”
When a job is missed, we do not replace it silently.
We make it visible.
We ask, “What happened” and “What is your plan.”
Then we let the child repair.
One week, a student forgot to water seedlings in the greenhouse.
A few plants wilted.
The student looked crushed. A teacher could have protected them from that pain by watering on the child’s behalf.
But the teacher chose a different lesson.
They walked back to the greenhouse together.
They watered the plants.
They talked about what seedlings need.
They made a new plan: a reminder note, a buddy check, and a shorter daily window that felt doable.
The child did not just learn about plants.
They learned they could face a mistake, repair, and try again.
That is self leadership in training clothes.
Agency is built in layers
We do not hand a child adult responsibility overnight.
We build agency the way we build strength: with small loads, repeated often.
For younger children, agency might look like:
Packing their own bag with a simple checklist.
Putting their plate in the sink and wiping the table.
Choosing clothes the night before.
For older children, agency might look like:
Emailing the teacher themselves when they have a question.
Planning their week on Sunday and bringing it to you for review.
Managing one household responsibility that actually matters, like making lunch for the family once a week.
The goal is not to make childhood into a boot camp.
The goal is to make responsibility normal, so self respect can grow.
Do not outsource the invisible work
The hardest part of growing up is not the task itself.
It is the invisible work around the task: noticing, planning, starting, and finishing.
This is the work AI can easily do for your child.
AI can write the plan, generate the checklist, and create the schedule.
If your child uses AI for those steps before they can do them themselves, they may look organized while staying dependent.
A simple way to protect development is to separate planning from polishing.
Let your child do the first, messy version of the plan.
Then, if you want, let AI help them make it clearer.
The rule is the same: the first draft belongs to the child.
A small ritual for self leadership
If you want to build agency without daily conflict, create a review rhythm.
Agency does not grow through nagging. It grows through reflection.
Once a week, do a ten minute “plan and review.”
Ask your child three questions.
“What went well this week?”
“What was hard?”
“What is one thing you want to do differently next week?”
Then help them make one small plan.
Not a perfect plan. A usable one.
The point is not the calendar.
The point is the identity shift.
I am the kind of person who plans, learns, and adjusts.
When your child says “I don’t want to”
Agency is not the absence of resistance.
Children resist responsibility because responsibility is effort, and effort is uncomfortable.
In the AI era, discomfort is optional. That is the trap.
When your child says, “I don’t want to,” do not argue about desire.
Name the reality and return the ownership.
“I get it. You don’t want to. And it’s still your responsibility. What’s your plan?”
This does two things.
It validates the feeling without giving the feeling authority over the outcome.
It also teaches a core life skill: we can do hard things without waiting to feel like it.
Avoid the control battle
Many parenting fights are not about the task.
They are about who is in charge of the child’s life.
When you nag, your child feels controlled and resists.
When your child resists, you feel disrespected and escalate.
The cycle is predictable, and it is exhausting.
The way out is to stop treating responsibility like a moral debate.
Treat it like a contract.
A contract is clear. It is calm. It is revisited.
A contract lets you say, “I am not going to argue about whether this matters. We already agreed.”
FAMILY EXPERIMENT
The Agency Contract
The goal of this experiment is to move one repeating conflict out of the daily emotional weather and into a simple agreement.
Pick one domain that causes friction in your home.
Examples: morning routine, homework, device use, chores, bedtime, or getting out the door on time.
Do this for two weeks. Keep it small enough that you can actually follow through.
Step one
Name the responsibility in one sentence.
Examples: “You pack your school bag the night before.” “You put your phone on the charger outside your room at 9:00.” “You start homework by 6:00.”
Step two
Write what your child owns.
Use simple, observable actions. Avoid vague words like “be responsible.”
Example: “I pack my bag, including water bottle, homework, and sports clothes. I check the list.”
Step three
Write what you own as the parent.
This is where you define support without rescuing.
Examples: “I will help you make a checklist.” “I will remind you once at 8:00 p.m.” “I will sit nearby for ten minutes while you start.”
Step four
Decide the consequence, and make it proportional.
The consequence should connect to the responsibility, not to your anger.
Examples: “If the bag isn’t packed, you wake up earlier to pack it.” “If the phone isn’t docked, it sleeps in the kitchen for the night.” “If homework isn’t started by 6:00, you lose screens after dinner until it is done.”
Keep consequences safe. Your job is to protect health and dignity, not to manufacture suffering.
Step five
Schedule a review.
Pick a time once a week. Ten minutes is enough.
In the review, ask: “What worked?” “What was hard?” “What do we want to adjust?”
A one page template
Write this on paper or in a shared note.
- The responsibility:
- What you own:
- What I own:
- What happens if it’s not done:
- Review time:
An example (morning routine)
Responsibility: You get yourself ready for school by 7:30.
What you own: Clothes are chosen the night before. You wake up when the alarm goes off. You eat breakfast and brush teeth.
What I own: I will wake you once. I will have breakfast available. I will not remind you again.
What happens if it’s not done: If you are not ready by 7:30, you go to school as is (within dignity and safety) and you go to bed fifteen minutes earlier that night.
Review time: Sunday after dinner.
One line that changes the tone
Before you start, say this to your child.
“I’m not doing this to control you. I’m doing this because I believe you can handle your life, and I want us to fight less.”
A version for AI use
If AI has become a point of conflict, make it the contract domain for two weeks.
Responsibility: You use AI in ways that protect your learning and integrity.
What you own: You do the first attempt yourself. You can explain your work out loud. You disclose when you used AI.
What I own: I will not spy on you. I will ask the Ownership Test questions. I will help you learn better prompts.
What happens if it’s not done: If you cannot explain the work, you redo it without AI, with support. If you hide your use, AI tools pause for a week while we rebuild trust.
Review time: Friday after school.
Closing: the goal is a child you don’t have to chase
When parents imagine “a responsible kid,” they often imagine a child who is easy.
But ease is not the point.
The point is ownership.
Ownership is what allows a teenager to navigate freedom without self destruction.
Ownership is what allows an adult to use powerful tools without becoming dependent on them.
In the AI era, your child will have endless assistance.
What they will not automatically have is self leadership.
That has to be trained, patiently, in the ordinary moments of home life.
You are not raising a résumé. You are raising a person who can steer.
A script for parents
“I’m on your team. I will help you plan. And I’m going to keep the responsibility with you, because I believe you can handle it.”
What to notice this week
One moment when you wanted to rescue, and what you did instead
One moment when your child owned a mistake without being pushed
One moment when AI helped your child learn, and one moment when it tried to replace ownership
A school question
“Where do students practice real responsibility and self leadership, and what happens when they drop the ball?”
A good school will answer with practices, not slogans.
Endnotes
[1] Ryan, R. M., & Deci, E. L. Self determination theory and the facilitation of intrinsic motivation, social development, and well being. 2000. Pages 68 to 78.
[2] Neubauer, A. B., et al. Daily autonomy supportive parenting, child well being, and parental need fulfillment. 2021. Locator: open access copy.
[3] Soenens, B., & Vansteenkiste, M. A theoretical upgrade of the concept of parental psychological control: Proposing new insights on the basis of self determination theory. 2010. Pages 74 to 99.
[4] Vigdal, J. S., & Brønnick, K. A systematic review of helicopter parenting and its relationship with anxiety and depression. 2022. Locator: Frontiers in Psychology, article 872981.
[5] Bruysters, N. Y. F., et al. Overprotective parenting experiences and early maladaptive schemas in adolescence and adulthood: A systematic review and meta analysis. 2023. Locator: open access copy.
[6] Seery, M. D., Holman, E. A., & Silver, R. C. Whatever does not kill us: Cumulative lifetime adversity, vulnerability, and resilience. 2010. Pages 1025 to 1041.
[7] Risko, E. F., & Gilbert, S. J. Cognitive offloading. 2016. Pages 676 to 688.
[8] Sparrow, B., Liu, J., & Wegner, D. M. Google effects on memory: Cognitive consequences of having information at our fingertips. 2011. Pages 776 to 778.
[9] Dahmani, L., & Bohbot, V. D. Habitual use of GPS negatively impacts spatial memory during self guided navigation. 2020. Locator: Scientific Reports, article 6310.
[10] Pew Research Center. About a quarter of US teens have used ChatGPT for schoolwork, double the share in 2023. 2025. Locator: January 15, 2025 short read.
[11] White, E. M., DeBoer, M. D., & colleagues. Associations between household chores and childhood self competency. 2019. Pages 176 to 182.
[12] Paley, B., & Hajal, N. J. Conceptualizing emotion regulation and coregulation as family level phenomena. 2022. Pages 19 to 43.
[13] Mueller, C. M., & Dweck, C. S. Praise for intelligence can undermine children’s motivation and performance. 1998. Pages 33 to 52.
[14] UNESCO. Guidance for generative AI in education and research. 2023. Locator: guidance document.