Imagine a Year 9 pupil who asks an artificial intelligence (AI) chatbot to explain the causes of the First World War. The answer is fluent, confident and warmly encouraging about her question. One date is wrong and one “historian” does not exist, and she has no way of knowing. That small moment is why AI literacy in schools has stopped being a nice extra. If pupils are going to use AI anyway, someone has to teach them how to doubt it.

In short: a secondary school in East Renfrewshire has appointed what is believed to be the first dedicated principal teacher of AI literacy in the United Kingdom (UK). Its youngest pupils learn to spot fake images, question chatbot answers and notice when a tool is flattering them. In England, a minister confirmed on 19 October that the refreshed curriculum will include AI knowledge, subject to consultation. A new international framework now sets out what pupils should learn. You do not need a new post to start.

What Scotland’s first AI literacy teacher is doing

The British Broadcasting Corporation (BBC) reported this month on Mearns Castle High School, where Jamie Laycock, a former principal teacher of biology, now leads AI literacy. Many UK schools squeeze an AI lead in alongside a day job. Mearns Castle made it a role after interest from parents and pupils, and Laycock will also train colleagues and develop materials for other schools.

All pupils in the first two years of secondary get AI literacy lessons. Older year groups meet it through citizenship and personal education classes, and senior pupils can take an elective. The message is blunt: “We can’t just hope AI goes away.”

The lessons cover hallucinations, bias and the habit of treating a chatbot like a person. Laycock is frank about the jargon. “I think hallucination is actually a really bad word, but it basically makes stuff up,” he told the BBC. He is equally direct about why older, better-read pupils have an advantage: “if you don’t have what we call ‘foundational knowledge’ it might be hard for you to figure out whether something is a hallucination or whether it’s genuine.”

The part I found most useful was on flattery. “It’s sycophantic,” Laycock said. “It always tells you you’re amazing, that your idea is great.” His worry is that young people will “get trapped and hooked into the technology and become over-reliant.” That is cognitive offloading by another name. One pupil, Aydin, said it is better to use AI “for how to get to the answer, rather than just letting it do it for you.” Another, Anna, came out of a lesson sorting real photos from AI-generated ones less sure she could tell them apart than when she went in. That is a good result.

The head teacher, Stuart Clark, draws a careful line: “acting on AI is not the same thing as advocating for it or promoting it.” He does not want schools to repeat “the same mistakes that society made with social media”.

The policy picture on both sides of the border

Scotland already has national direction. The Scottish Government’s guidelines and guardrails for AI in schools, published in March, say pupils “should learn about, understand and use AI, in appropriate contexts” as part of their entitlement to skills for learning, life and work. The same document insists that AI “can never and should never act as a replacement for human interaction within classrooms” and that teachers decide “whether, when, and how to use AI”. Pupils and teachers should both be able to critique and challenge what these tools produce.

England is moving more slowly. Georgia Gould, a Minister of State at the Department for Education, answered a written parliamentary question on 19 October. Following the Curriculum and Assessment Review, she said, “our refreshed curriculum will equip pupils with essential technological and digital skills, including the knowledge needed to engage with artificial intelligence (AI) safely, responsibly and in an age-appropriate way.” She added: “All proposed content will be subject to public consultation.” That is a promise, not a scheme of work. Pupils using chatbots this term cannot wait for it.

A framework for AI literacy in schools

The clearest structure I have seen is the AI Literacy Framework for Primary and Secondary Education (AILit Framework), a joint project of the European Commission and the Organisation for Economic Co-operation and Development (OECD). The final version, published in June after a draft and an international review, also feeds into the Programme for International Student Assessment (PISA) 2029 assessment of media and AI literacy. The full framework document is free to read. It is non-binding, which is a strength: it describes what pupils should be able to do and leaves the teaching to us.

It groups 19 competences into four domains. Engage with AI (seven competences) is about being a critical participant in a world shaped by AI. Create with AI covers using it as a creative partner “while maintaining human agency”. Manage AI asks pupils to “divide work intentionally between humans and AI”. Shape AI is about improving AI systems so that they reflect human values. Each competence comes with basic, intermediate and advanced expectations and short classroom scenarios.

One line from the framework deserves a place on every staffroom wall: “AI literacy itself does not rest on a single educator” or a single subject. That matches what Laycock said about foundational knowledge. A pupil can only catch a wrong answer about photosynthesis if she knows some biology.

Four lessons you could run this term

None of these needs a specialist. They draw on the BBC report and the framework’s scenarios, and the classroom details are suggestions, not tested lesson plans.

Real or generated?

Mearns Castle pupils sorted real images from AI-generated ones and found it much harder than they expected. You could try the same in a geography or art lesson with ten pictures of coastlines, half of them generated. Ask pupils to commit to an answer and give a reason, then reveal the results. The point is not to train a perfect eye. It is to show that confidence and accuracy are different things, which links neatly to our earlier piece on school photos and AI.

Accept, revise or reject

One of the framework’s competences asks learners to “evaluate whether AI outputs should be accepted, revised or rejected.” Its basic scenario has pupils compare an AI solution to a maths problem with the teacher’s method. In a Year 10 history lesson, you might give pairs a chatbot summary of an event they have just studied and ask them to mark each claim as keep, fix or bin, with a source for every fix.

The four corners

In one intermediate scenario, pupils move to corners of the room labelled “AI Only,” “AI-Supported,” “Human Only,” or “Not Sure” as the teacher reads out tasks. Try it before a piece of coursework. Spelling check, research plan, personal reflection: where does each belong? The arguments in the “Not Sure” corner are where the learning happens, and it fits well with agreeing rules on AI cheating in schools before an assignment begins.

The flattering machine

This one is mine, prompted by Laycock’s point about sycophancy. Ask a chatbot to rate a deliberately weak paragraph and note how kind it is. Then ask it to be honest. Pupils see that the tool’s tone shifts with the prompt, and they start to read praise from a machine with healthy suspicion.

The most useful thing a pupil can learn about a chatbot is that its confidence tells you nothing about whether it is right.

Who should teach it?

My view is that every school needs a named person, but AI literacy cannot live only in computing or a single assembly. The Mearns Castle model works because it has dedicated time and then reaches out into citizenship, personal education and other subjects. Most schools cannot fund a new post. They can give someone a few hours, put a short AI literacy strand into personal, social and health lessons, and ask each department which framework scenario suits its subject. The framework’s own scenarios include checking restaurant recommendations for a travel plan and verifying an AI account of a historical event against trusted sources, so there is something for almost everyone.

Start with staff. The Scottish guidance says teachers as well as pupils “must be supported to develop an appropriate understanding of how AI works, its impacts, and its risks.” A twenty-minute briefing where colleagues try the flattering machine exercise on their own writing would do more than another policy document.

Is your school teaching AI literacy yet, and who has picked it up? I would love to hear which lessons have worked with your classes, so please share them in the comments.