Raising Children in the Age of AI: Why Judgement Matters More Than Ever

Artificial intelligence is changing the way we find information, solve problems, produce work and make decisions. For our children, it will increasingly be part of the environment in which they learn, study and eventually work. Yet much of the conversation about preparing children for AI focuses on which technical skills they should acquire or which careers might survive automation.

I think there is a more fundamental question. If technology can increasingly generate answers, how do we help children develop the judgement to know which answers deserve their trust? An answer can be articulate, convincing and completely wrong. It can also be technically accurate while overlooking something important about the situation in which it will be used.

As a parent, this is the aspect of AI that interests me most. I cannot predict what my son will need to know professionally in twenty years. But I can see why the ability to evaluate information, recognise uncertainty, question assumptions and make considered decisions may become increasingly important. This sits within a wider question I have been exploring about What it means to raise capable children not simply children who achieve, but children increasingly able to think, decide and navigate for themselves.

Why Raising Children in the Age of AI Requires a Different Approach to Judgement

For much of modern education, finding and retaining information has been central to demonstrating knowledge.

That remains important. Children need substantial knowledge to understand the world and evaluate what they encounter. But generative AI introduces a different challenge.

A child can now receive a plausible explanation of almost anything within seconds. They may not know whether the answer is accurate, whether the evidence is reliable or whether important qualifications have been omitted.

The problem isn't simply that AI sometimes makes mistakes. It is that the quality of an answer's presentation can be mistaken for the quality of its reasoning. A confidently expressed answer may feel authoritative even when it deserves considerable scepticism.

Children therefore need to learn to distinguish between receiving information and deciding whether that information is trustworthy. This distinction is becoming central to discussions about AI literacy.

The OECD and European Commission's 2026 AI Literacy Framework for primary and secondary education identifies the ability to critically evaluate AI outputs and make informed decisions as important elements of AI literacy.

It is not enough to know how to use a tool. Children also need to understand its limitations.

Judgement Is More Than Critical Thinking

We often use the phrase critical thinking when discussing the skills children will need in the future. But judgement is slightly different.

Critical thinking helps us examine an argument, identify assumptions and evaluate evidence.

Judgement involves deciding what to do with that information.

A child might understand that an AI-generated explanation contains uncertainty. They must then decide whether to accept it, investigate further, seek another perspective or refrain from making a decision.

Judgement therefore involves more than detecting factual errors. It includes recognising when there isn't enough information, understanding the consequences of being wrong and knowing when another person's expertise is needed.

Consider a child researching a historical event.

An AI tool might provide a coherent account of what happened. The child may need to ask whether the explanation reflects the perspectives of everyone involved, whether the sources are credible and whether the account leaves out important context.

Now consider a different situation: a teenager asking an AI tool for advice about a disagreement with a friend. The challenge is no longer primarily factual. The tool doesn't know everything about the relationship, the people involved or the consequences of following its advice. The teenager needs to recognise that limitation.

Good judgement means understanding not only whether an answer sounds reasonable, but whether the system providing it is in a position to know.

Why Knowledge Still Matters When AI Can Provide Answers

There is a temptation to conclude that children will need to memorise less because information is so readily available. I think that argument is too simplistic.

You cannot evaluate an answer effectively if you know almost nothing about the subject.

A child who understands history, science, literature or mathematics has a foundation against which to assess new information. That foundation begins long before children encounter AI. Reading has been one of the things I have valued from my son's earliest years, not simply for literacy but for the language, knowledge and ideas it gradually puts at a child's disposal. I explore that separately in The Importance of Reading to Your Child From When They Are a Baby.

Without that foundation, even identifying an implausible claim becomes difficult.

This is an important distinction between using AI to extend knowledge and using it to replace the process of acquiring knowledge.

The OECD's 2026 discussion of generative AI in education highlights the difference between producing better work with AI assistance and developing understanding that can be demonstrated independently. Its analysis cautions that improvements in AI-assisted performance do not necessarily translate into equivalent learning gains.

That distinction matters for parents.

A polished piece of homework may tell us relatively little about what a child has understood if a significant part of the reasoning was performed by a tool.

The question becomes whether the child can explain, defend and apply the ideas independently.

What Research Tells Us About AI and Children's Thinking

The research is developing quickly, but it is important not to overstate what we know.

Much of the available experimental evidence concerns university students rather than primary-school children. We should therefore be cautious about drawing direct conclusions about younger children's long-term cognitive development.

Nevertheless, the findings raise questions worth considering.

A 2026 systematic review in Computers and Education: Artificial Intelligence examined 67 empirical studies published between 2022 and 2025. The researchers found that AI could support critical and creative thinking when used within structured learning activities that encouraged reflection, reasoning and evaluation. However, less structured use could encourage students to transfer too much of the thinking process to the technology.

The researchers emphasised that outcomes depended heavily on how AI was integrated into learning, rather than simply whether it was used (Reference). Another 2026 systematic review, published in Thinking Skills and Creativity, similarly found potential benefits for analytical and reasoning skills when AI was appropriately integrated into teaching.However, the authors reported that evidence for improvements in underlying critical-thinking dispositions, such as truth-seeking and openness, remained inconsistent. (Reference).

That distinction is fascinating.

A child or student may perform a particular analytical task more successfully with AI assistance without necessarily becoming more inclined to question information independently.

The tool may improve the immediate result without producing the same change in the person's underlying habits of thought.

For me, this is where the discussion becomes particularly relevant to parenting.

We shouldn't simply ask whether children can use AI effectively.

We should ask whether they are developing the intellectual habits that allow them to evaluate what it produces.

The Difference Between Getting an Answer and Understanding It

Imagine a child is struggling with a mathematics problem.

An AI tool can explain the solution, identify the error or complete the calculation.

All three may be useful in different circumstances.

But they do not create the same learning experience.

If the tool immediately provides the answer, the child may finish the task without understanding why the method works. If it identifies an error and asks the child to reconsider their reasoning, it may support learning more effectively.

This is not an argument against using AI. It is an argument for being clear about which part of the thinking process we want the child to retain.

The OECD's 2026 Digital Education Outlook discusses precisely this tension: generative AI can improve immediate task performance while potentially reducing the effort required for deeper learning when it is used without a clear educational purpose.

The implication is not that every difficult task must be completed without assistance.

It is that assistance should serve learning rather than quietly replace it.

Children Need to Recognise Uncertainty

One of the most important aspects of judgement is knowing when we do not know enough.

Adults struggle with this too.

We often prefer a clear answer to an ambiguous one, particularly when a decision feels urgent.

Generative AI can make this more complicated because it is capable of presenting uncertain information in remarkably fluent language.

A child might ask a question and receive a detailed explanation without realising that the answer involves interpretation, incomplete evidence or genuine disagreement. Learning to recognise uncertainty therefore becomes an important part of AI literacy.

Children should gradually become comfortable asking whether a question has one correct answer, whether the evidence is sufficient and whether different interpretations are possible.

They also need to recognise that some questions are better answered by people with relevant expertise or personal knowledge.

This is also one reason I value children's encounters with art and culture. Not everything they encounter has a single correct interpretation, and learning to look, question and form a response has value of its own. I explore that more in The Benefits of Arts and Culture for Children.

UNESCO's AI competency framework for students places particular emphasis on a human-centred understanding of AI, ethical considerations and the critical evaluation of AI-generated solutions. That is a useful reminder that judgement is not simply a technical skill.

It also involves understanding the appropriate role of technology in human decisions.

How Parents Can Help Children Develop Better Judgement About AI

I don't think parents need to become experts in artificial intelligence to help their children think more carefully about it. What matters is how we approach questions, evidence and uncertainty.

For younger children, much of this can happen without using AI at all. When a child encounters a surprising claim in a book, conversation or online video, we can discuss how someone might know whether it is true. As children become older and begin using digital tools more independently, those conversations can become more specific.

Instead of immediately accepting an AI-generated answer, we can encourage them to examine:

What evidence supports this?

How could we check it?

What might be missing?

Does this answer make sense given what you already know?

Is this something a computer can reliably decide?

These aren't questions that need to be asked every time a child uses technology. That would quickly become exhausting.

They are ways of establishing a habit: an answer is something to consider, not necessarily something to accept.

For children using AI for schoolwork, another useful distinction is between asking a tool to explain something and asking it to complete the thinking on their behalf.

A child might ask AI to explain a concept in a different way, suggest questions for revision or identify possible weaknesses in an argument they have already developed. Those uses preserve more of the child's responsibility for understanding the material.

Age matters, however. Younger children should not be encouraged to use general-purpose AI tools independently without considering the service's age restrictions, privacy arrangements and school guidance.

UNESCO's guidance on generative AI in education recommends an age-appropriate, human-centred approach to its use.

The Risk of Confusing Confidence With Accuracy

One of the more subtle challenges of AI is that it can sound remarkably certain.

A child who has spent years learning that confident explanations usually come from knowledgeable adults may find it difficult to distinguish between a reliable explanation and one that merely sounds reliable.

This is particularly important because AI-generated information is not always obviously wrong.

An answer may contain several correct facts alongside one important mistake. It may present a reasonable interpretation without acknowledging alternatives. Or it may offer advice that is sensible in general but inappropriate for a particular individual.

Judgement requires recognising these distinctions.

The OECD has highlighted the importance of ensuring that young people understand the difference between AI-generated suggestions and established facts, particularly as they become more autonomous users of technology.

I think this is an important consideration for families because children increasingly encounter information outside environments where a teacher or parent has already evaluated it.

They need ways of assessing credibility for themselves.

AI Can Support Judgement If We Use It Deliberately

It would be a mistake to frame AI entirely as a threat to children's thinking.

Used appropriately, it may provide opportunities to examine competing explanations, test an argument or identify weaknesses in reasoning.

A student might ask an AI system to challenge their interpretation of a historical event.

Another might use it to generate objections to an essay argument before deciding which objections are persuasive.

The important difference is that the student remains responsible for evaluating the responses.

Research on AI-supported learning increasingly suggests that the design of the interaction matters. A 2026 review of 67 empirical studies found more promising outcomes when AI was used within structured activities involving reflection, argumentation and evaluation, rather than simply as a convenient answer generator. That suggests a useful principle.

AI should give children more to think about, not remove the need for them to think.

The distinction may become increasingly important as the technology becomes more capable and easier to use.

What Should Children Still Be Able to Do Without AI?

This is the question I keep returning to. Not because I think children should avoid technology, but because we need to be clear about which abilities matter independently of it.

I want my son to be able to read something and decide whether he understands it.

To recognise when an argument is weak.

To distinguish evidence from opinion.

To notice when important information is missing.

To acknowledge uncertainty without feeling that he must immediately resolve it.

And to understand that some decisions involve values, relationships and consequences that cannot be reduced to a technically convincing answer.

None of these abilities makes AI unnecessary. They make it more useful.

A child who can evaluate information, recognise limitations and think independently is better placed to benefit from powerful tools without becoming entirely dependent on their conclusions.

Raising Children for a Future We Cannot Predict

I don't know what my son's professional life will look like. I don't know which jobs will exist, which qualifications will matter most or how extensively

AI will and is becoming integrated into everyday decisions and work environments. And I am sceptical of anyone who claims to know with certainty.

But I do think we can identify an important educational responsibility. As information becomes easier to generate, children will need to become increasingly thoughtful about what deserves their attention and trust. They will need knowledge, certainly. They will also need the ability to evaluate claims, recognise uncertainty and make decisions that take account of context and consequences.

There is also a danger that anxiety about AI simply becomes another pressure to add to childhood, another skill children need to acquire, another area in which they need to get ahead. I've written separately about Raising children in Competitive Environments and why I think we need to be careful about allowing preparation for the future to consume the childhood they are living now. These are not entirely new human abilities. What is changing is the environment in which they must be exercised.

What is changing is the environment in which they must be exercised. We are entering a world in which a child may receive an extraordinarily sophisticated answer before they have fully understood the question.

My concern isn't that AI will know more than our children. It almost certainly will, in many domains. My concern is whether they will develop the judgement to recognise when its answers are useful, when they are incomplete and when they should not be trusted.

Judgement is not something children acquire because we tell them to think critically. It develops through repeated opportunities to weigh information, make decisions, discover that they were wrong, reconsider and decide again. AI doesn't change the importance of developing judgement. It changes the conditions in which children will learn to exercise it.

That, to me, is one of the most important questions in raising children in the age of AI.

Shanti

The Kensington Diary