Six Commitments for Childhood in the Age of AI

Six Ways to Grow Agency

Helping children think, choose and stay capable in the age of AI

The Parenting Blind Spot asks parents to notice the changing relationship children are developing with AI, including what they ask it, why they turn to it, what they trust it with and what they gradually choose to hand over.

Recognising the Blind Spot leaves us with a practical question:

How can we help children benefit from increasingly capable AI while continuing to develop their own capabilities?

AI literacy is an important part of the answer because children need to understand the technology they are growing up with, learn to question its outputs and recognise both its possibilities and its limitations. But understanding alone doesn’t necessarily give a child agency.

In Growing up in a Digital World, BSI describes agency as the real objective. I think that gives us a useful way to think about where we want children to arrive. Agency develops throughout childhood as children gain the capability, confidence and opportunity to make increasingly intentional choices about their own lives. In a world containing AI, some of those choices will concern when to use technology, how to use it, what to believe, what to delegate, when another person is the better place to turn and when to disengage. For a child, perhaps the experience can be expressed much more simply:

I can. I choose.

These six ideas give families practical opportunities to help children practise that agency while adult guidance is still available.

1. Explore Together

Access, yes. Unguided access, no.

AI literacy should come before AI independence.

Children will encounter AI, experiment with it and discover what it can do. Exploring alongside them gives parents an opportunity to understand that world too, while children can begin recognising both the possibilities and limitations of the technology.

This doesn’t mean permanently supervising every interaction. As children’s understanding and judgement develop, parents can gradually step further back and give them greater responsibility for their own choices.

Independence has somewhere to grow from.

For the child: I can explore this, and I know when to ask for help.

Try it: Ask Together

Choose something you are both curious about and ask AI together, then talk about the response. What was useful? What surprised you? What might be missing? Is there anything you would want to check? Could you ask the question differently?

The exercise makes conversation part of using AI from the beginning.

2. Question What Sounds Convincing

AI’s fluency isn’t truth.

Teach children to question, check and challenge.

AI can produce language that sounds remarkably confident, even when an answer is misleading. Children therefore need experience of noticing that confidence, asking what supports an answer and checking important claims before deciding what they believe. Over time, questioning can become part of their own judgement rather than simply a rule imposed by an adult.

For the child: I decide what deserves my trust.

Try it: Catch the Confidence

Ask AI something together and look for parts of its answer that sound particularly certain.

Choose one claim and investigate it elsewhere, then talk about what made the original answer sound believable before you checked it.

The purpose is to develop a feel for the difference between sounding confident and earning trust.

3. Keep Thinking Yours

Augmentation before automation.

Use AI to strengthen capability, not replace it.

AI can provide a finished answer, although sometimes a clue, explanation, challenge or piece of feedback will be more useful to a developing child.

A child struggling with homework could ask AI to complete it, or they could ask for help organising their thoughts, understanding a difficult concept or identifying where their reasoning has gone wrong.

The question I keep coming back to is simple:

Does using AI leave the child more capable afterwards?

This is what I mean by augmentation before automation.

For the child: I can get help without handing over my thinking.

Try it: Help Me, Don’t Do It For Me

Encourage your child to make an attempt first and then ask AI for the smallest amount of help needed to move forward.

‘Give me one clue.’

‘Ask me a question that will help me work this out.’

‘Tell me where my reasoning went wrong without giving me the answer.’

Afterwards, ask: Could you do more yourself now than you could before?

4. Choose Where You Turn

AI’s attention isn’t love.

Simulated attention isn’t the same as human care.

Conversational AI is easy to talk to, always available and capable of responding patiently to questions children might find uncomfortable asking another person. That can sometimes be helpful, although children also need opportunities to think about who or what a question belongs with.

AI may be useful for explaining photosynthesis or helping to practise French. A friendship problem may benefit from someone who understands the people involved, while a difficult feeling may need a trusted person who can listen, notice, care and respond to what happens next. Sometimes the child may simply need enough space to discover what they think for themselves.

For the child: I choose where I take my questions.

Try it: Who Would You Tell?

Think of different questions together and decide where you might take each one:

MYSELF | ANOTHER PERSON | AI | MORE THAN ONE

Begin with easy examples and gradually make them more difficult.

There doesn’t always need to be one correct answer. The useful part is asking the child to explain why they made their choice.

5. Stay in the Conversation

Parents must stay close while children learn to choose.

Every question can begin a conversation. Conversations can build trust.

Today the question might be homework. Tomorrow it could be:

“Why don’t my friends like me?”

Our role isn’t to provide faster answers than AI. We can’t. We can offer something different: knowledge of the child, lived experience, judgement, accountability, reciprocity and care. Staying in the conversation also means allowing children to disagree with us, make choices, get things wrong and gradually need less of our guidance. Agency cannot develop if every choice is made for them.

For the child: I can make choices, and I know there are people I can turn to.

Try it: Who’s in Charge?

Put everyone’s phones face down for twenty minutes. Parents play too.

Whenever somebody notices the urge to pick theirs up, simply say ‘Pull’ and notice what caused it. Perhaps it was a notification, boredom, habit, curiosity, fear of missing something or a genuine need to use the phone. Then choose whether to pick it up.

There’s no penalty for doing so because agency doesn’t mean not using the phone. It means becoming better at noticing the moment when a choice is being made.

Notice → Pause → Choose

BSI’s report records one young person saying:

‘I wish I could put my phone down.‘

Sometimes practising agency begins in the small space between influence and action.

6. Decide What Stays Human

Not everything that can be handed to AI should be.

AI will increasingly be able to write, create, advise, recommend, remember and act on our behalf. Children growing up alongside these systems will therefore face choices previous generations rarely had to make.

What am I happy for AI to do?

Where would I like it to help me?

What do I want to continue doing myself?

Which experiences do I value because another human being is involved?

Which decisions do I want to remain mine?

There will be no single permanent set of correct answers. They will change as children mature and as the technology develops. Learning to ask the questions is part of learning to exercise agency.

For the child: I decide what I give to technology and what I keep for myself.

Try it: What Stays Human?

Choose some ordinary activities and decide whether you would:

DO IT MYSELF | DO IT WITH AI | LET AI DO IT

Try writing a birthday message, choosing a present for a friend, solving homework, creating a picture, deciding how to apologise, planning a journey or working out what to do when somebody is upset. Different people may make different choices. Ask everyone to explain theirs.

Deciding what to delegate is itself a skill.

What Does Growing Agency Look Like?

No point marks when a child suddenly acquires agency, and these six ideas are neither a test nor a scorecard. Agency develops through experience.

We may see it when a child tries before automatically asking AI, questions an answer that sounds certain, asks for a clue rather than completed work, recognises that another person would be a better place to take a question, notices an automatic reach for a device, or explains why they want AI’s help with one thing but would rather do another themselves.

Those may seem like small moments, but they give children repeated experience of something important:

I can. I choose.

Parents Cannot Carry This Alone

Helping children develop agency cannot all rest on the shoulders of parents and children. BSI makes an important distinction here:

“Ultimately, developers, platforms and service providers should bear primary responsibility for ensuring that products are safe by default.”

BSI, Growing up in a Digital World

Children cannot reasonably be expected to overcome technologies deliberately designed to maximise engagement simply because we’ve taught them about algorithms or told them to exercise self-control. Children need safe environments in which meaningful choices are actually possible.

Parents have a different role.

We can give children opportunities to practise making those choices while we are still close enough to help.

Protect and prepare, rather than simply prohibit.

Children growing up today may have access to forms of assistance more capable than anything previous generations experienced. We should want them to benefit enormously from that capability while still reaching adulthood able to think through a difficult problem, question something persuasive, create something of their own, make a decision, approach another person, change their mind, recognise when they need help and choose when technology is the right tool.

Perhaps that is what growing agency ultimately feels like:

I know this technology can do extraordinary things.

I also know that I can.