AI literacy is becoming essential. Dr Martin Saunders, CEO of EdShed, argues that primary schools needn’t teach children to use AI but should teach them to think critically about the technology shaping their world and how it works.
For many primary-aged children, their first encounter with Artificial Intelligence comes long before their first lesson about it.
Algorithms already influence their lives, from recommendations determining which videos they see to Generative AI tools creating text, images and answers.
While we’re still debating when children should start learning how it works, in reality, the question is no longer whether children are old enough to learn about AI, but whether we can afford to wait.
AI literacy as part of everyday literacy
Literacy has always involved understanding and interpreting information. Today, more than ever, children also need to understand how information reaches them and why they are being shown it. AI education isn’t primarily about training future programmers; it’s about creating informed citizens.
AI is changing not just how information is created, but how it is selected, presented and consumed. Recommendation systems can influence what children encounter online, while Generative AI can produce convincing text, images and answers that are not necessarily accurate.
This creates a new challenge for education. Children need to develop the ability to question where information comes from, recognise that technology is not always neutral and understand why a confident-sounding answer might still be wrong. Just as we teach children to assess the reliability of what they read, we now need to teach them to question the systems increasingly responsible for delivering that information. Understanding AI matters more than simply knowing how to use it.
Children don’t need to understand the technology at an advanced level
Some will question whether primary school is too young, but I’m not suggesting that seven-year-olds need to understand neural networks or the mathematics behind Machine Learning.
Instead, the focus should be on introducing fundamental concepts through simple, age-appropriate activities. Children can learn how algorithms work, explore how AI systems identify patterns in data and begin to understand why the information generated by AI isn’t always reliable. None of this requires advanced technical knowledge or direct access to AI tools.
An introduction to algorithms could be as simple as asking pupils to write step-by-step instructions for a classmate to complete a task, demonstrating how computers follow rules and why precise instructions matter.
To explore how AI learns from data, children could sort pictures into categories and consider what would happen if some examples were missing or incorrectly labelled, introducing the idea that the information used to train a system influences its results.
Older primary pupils could compare an AI-generated explanation with a trusted source, identifying inaccuracies and discussing why an answer that sounds convincing isn’t necessarily correct.
AI isn’t magic
Sophisticated AI can appear remarkably human. Chatbots respond conversationally and confidently, creating the impression that they genuinely understand children or can automatically be trusted.
AI systems aren’t independent sources of truth. They are created by people, shaped by the data used to train them and influenced by human design choices. They can reproduce biases, generate inaccurate information and present uncertain answers with unwarranted confidence.
The objective isn’t to make children suspicious of technology, but to encourage a healthy question: ‘How does it know that?’
The alternative isn’t ‘no AI education’
The choice for schools is between helping children develop an informed, age-appropriate understanding of the technology and leaving them to make sense of it through their own experiences with apps, search engines, online content and their peers.
Structured education offers an opportunity to build curiosity, critical thinking and confidence in a safe environment, without requiring children to engage directly with potentially unsuitable AI tools. We should be encouraging children to ask questions about AI, not simply teaching them to accept or use its outputs.
Don’t make AI another bolt-on subject
Schools are already under enormous curriculum pressure, so AI literacy shouldn’t necessarily mean squeezing another lesson into the timetable.
The priority should be to embed fundamental AI concepts into existing learning. Computing provides a starting point for exploring algorithms and data, while English offers opportunities to question information sources and distinguish fact from fiction. Maths can introduce the importance of patterns and data, while Science and History can help children understand evidence, reliability and the need to test claims. Personal, social, health and economic (PSHE) education and online safety lessons also provide opportunities to explore AI’s wider impact.
Crucially, this needs to be a progressive approach rather than a one-off lesson. Younger pupils can begin by understanding how digital systems follow instructions, while older primary pupils explore how AI can recognise patterns, alongside other issues such as bias, misinformation and the limitations of AI-generated answers.
Those of us developing educational technology have a responsibility here too. It’s not enough to build tools that make learning easier or more engaging. We also need to consider how technology can help children develop the understanding and judgement to navigate an increasingly digital world. The priority should be to support educators, not simply introduce more technology into classrooms.
Successful implementation will also depend on teachers having the confidence, training and resources to introduce these ideas appropriately. We cannot expect schools to prepare children for an AI-enabled world without ensuring educators themselves have the support to do so.
A long-term digital skills issue
Much of the debate around AI skills focuses on the workplace and how businesses can equip employees to use emerging technologies. But digital confidence and critical thinking don’t begin when someone enters employment. They develop much earlier and the foundations for understanding AI should be laid during primary education.
AI will continue to evolve and the technologies today’s primary pupils encounter may look very different by the time they enter the workplace. But the ability to question information, recognise limitations and exercise independent judgement will remain essential.
The challenge for education isn’t to predict which AI technologies children will use in 10 or 20 years. It’s to equip them with the skills to adapt to technologies that don’t yet exist, while retaining the confidence to question what those technologies tell them.
We wouldn’t expect children to navigate the Internet without teaching them about online safety and nor should we expect them to navigate an increasingly AI-driven world without helping them understand the technology shaping it.
AI is already part of children’s lives; understanding it should be part of their education.
