Judgement is the skill: teaching digital resilience from primary upward
Knowing who made a piece of content, why, and whether to trust it is harder in the age of AI, and it matters younger. A case for teaching truth discernment on purpose, built from the primary years up.
The most useful four words in my primary school this year have been "how do you know?" A class that has learned to put that question to a search result, an image, or a confident answer from a chatbot is doing something harder than finding information. They are judging it. Building that habit is a large part of my job as the person responsible for how our primary phase approaches AI.
The skill underneath the skills
That habit has a name: judgement. Knowing who made a piece of content, why they made it, and whether it deserves your trust. It is the skill that holds up when the technology changes, and AI has made it harder to exercise.
In the first piece in this series I made the case that human intelligence matters more as AI spreads through our classrooms. Judgement is where that begins in practice, and it now begins earlier than it used to.
A primary classroom in 2026 is a place where a convincing fake image, a confident wrong answer and a source that was never real can all turn up in the course of ordinary work. A child asks a chatbot a question and gets a fluent, well-formatted paragraph that happens to be wrong. Another finds an image that looks exactly like a news photograph and was generated on a phone. None of this is exotic any more. It arrives at younger and younger ages, which means the judgement we used to leave until secondary school has to start sooner.
The three questions of judgement
Who made this?
Why did they make it?
Does it deserve my trust?
The world is moving the same way
This is why I read the programme for UNESCO's Digital Learning Week in Paris this month with some interest. The theme is "Education in the age of AI: facts, frictions, frontiers," and education ministers are expected to adopt a joint statement on protecting education as a common good. The middle word, frictions, is the honest one. The friction I would put near the top is truth discernment.
Policy is moving the same way. England's updated safeguarding guidance for schools took effect this September, and the wider direction of child-protection thinking now takes for granted that AI-generated and AI-altered images are part of a child's world. I sit outside that jurisdiction, in Astana, but the signal travels: once the people who write safeguarding rules assume synthetic content is everywhere, teaching has to assume it too.
The judgement that will be measured at fifteen is built in the years before it, lesson by lesson, from primary upward.
There is a longer horizon too. From 2029, PISA, the OECD's international study comparing how fifteen-year-olds perform across school systems, is adding a Media and AI Literacy assessment, testing whether they can engage with digital and AI-mediated content critically and responsibly.
Teaching it on purpose
For us it means treating truth discernment as content we teach and progress, the way we would any other skill. It builds across the phases.
How judgement is built, phase by phase
Is this real or made up? Who is telling me this, and how do they know?
Check a claim against a second source. Notice when an image or a number is doing persuasive work.
Ask what a tool was trained on, and what it might be leaving out.
Done well, it is a genuine progression built across the years, so a leader can see how the judgement develops and where it sits in each one.
None of it needs a new subject or a screen in every hand. It needs the questions planned deliberately, taught by staff who feel equipped to teach them, and revisited often enough to become a habit.
The check
The check I am running with my own team this term is a plain one: in each year group, can we name the lesson where a child is taught to ask whether something is true? Where we cannot yet, that is where we start.
