The conversation around artificial intelligence (AI) in schools is moving fast. Just a few years ago, the focus was on limiting its risks, preventing plagiarism and cheating. But time and tech wait for no man, and the question now has shifted to how can schools ensure pupils are AI literate?
For Ceri Jones, Headmaster at Caterham School in Surrey, there is “a real urgency” to the issue. “We keep talking about a world that is coming to us,” he says. “The reality is that the world is already here.”
Educators, he believes, are recognising that future success will depend less on grades or university destinations than on how young people respond to uncertainty and disruption.
For Caterham, that meant developing an explicit curriculum focused on “future-ready skills”. The school’s EDGE curriculum – explore, develop, grow, evolve – gives pupils, from the first year of prep to the upper sixth, dedicated digital innovation lessons, developing technological confidence alongside problem-solving, teamwork, entrepreneurship and self-reflection.
This September, Caterham is launching a new GCSE-equivalent Ethical AI & Entrepreneurship qualification in partnership with Nottingham University Business School and Enactus UK, the social entrepreneurial education charity. Validated by business leaders and education specialists, the course combines entrepreneurship, ethical AI and social impact, with pupils developing AI-assisted solutions to real-world challenges under the guidance of university mentors. Successful students will receive a GCSE-equivalent qualification that can be included on university applications.
“This isn’t just about teaching kids how to make money,” says Jones. “It’s about teaching them how to have an impact and change the world for the better.”
“Our job as educators is to create a sense of optimism”
Many pupils, he says, feel overwhelmed by global challenges. “Our job as educators is to create a sense of optimism, to make them believe they can solve problems.”
The new qualification has also struck a chord with parents. When Caterham explained that pupils would study ten GCSEs plus the new course rather than 11 GCSEs, Jones admits the school was unsure how families would respond. “But it went down really well. A number of parents just said: ‘Thank God, this is exactly what schools should be doing.’”
Alongside curriculum changes, Caterham has invested heavily in its own AI technology. Rather than relying on commercial platforms, it recruited machine-learning engineers to develop AI tools designed specifically for schools.
“We were determined to ensure that what we put in front of our pupils, in terms of AI or enhanced technology, is independent by educational purpose and pedagogy. I think, too often, the off-the-shelf products you buy don’t have pedagogy safeguarding or ethical use at their heart,” says Jones.
Its education technology company, Sphinx AI, created RileyBot, an AI learning assistant now shared for free with local state schools. Rather than completing work for pupils, RileyBot guides them through problems much like a teacher would.
“It’s about developing pupils’ agency around AI,” says Jones. “The reality is pupils are experimenting with AI whether we like it or not. My view is: let’s help them do it safely, securely and productively.”
AI is now embedded across school life. Jones describes pupils using large language models (LLMs) to generate code and build websites, while Year 7 pupils recently created an app to help new students navigate the transition from primary school, using school data to build an interactive school map.
But he is equally clear that traditional disciplines remain essential.
“We’re also teaching ethics, philosophy and critical thinking,” he says. “That’s also part of the AI education: you know – what is truth, what is reality?”
“We can no longer just teach kids to pass exams in the expectation that they’ll go to university and get into the world of work,” says Jones. “Whilst that pathway will still be there, I suspect, for a long time, I think there’s this other pathway emerging and I want all pupils to know they have the opportunity to build their own futures.”
For the Sherborne schools in Dorset, AI ambition is spreading beyond the school walls and into the town itself. This year, the boys’ school is marking the centenary of the arrival of its famous alumnus, Alan Turing, the father of modern computing. This month, the newly formed Sherborne Schools Group – Sherborne Boys, Sherborne Girls and Sherborne Prep – will formally unite under one brand. Central to its vision is the Sherborne Turing Project, a multimillion-pound initiative to transform this somewhat sleepy west country town into a hotbed for digital innovation and education.
The project includes the creation of The Turing Centre, an innovation, AI and entrepreneurship hub, alongside the inaugural Turing Conference this October, hosted by the school. Organisers hope it will become an annual gathering of leaders from technology, academia and industry.
Alastair Poulain, Deputy Head (Co-Curricular) at Sherborne Prep says the vision was driven by the town itself. A key aim is raising “the aspirations and ambitions” for all the young people of the town; outside of Sherborne’s school walls, opportunities are in short supply for many.
The conference will give Sherborne pupils and those from neighbouring schools “a chance to meet and hear from the great and the good in AI and tech”, he says. Speakers include AI luminaries such as Kerensa Jennings, Suki Fuller, Jamie Bartlett and Alan Turing’s nephew, Dermot Turing, while organisations including Bletchley Park, GCHQ and BAE Systems will take part in an afternoon tech festival.
The aim for the Turing Centre is to create a bold commercial space for Sherborne pupils and the people of Sherborne, providing AI education, creative labs, a technology incubator and start-up support, with ambitions to attract businesses and establish a university presence in the town. “We’re a bit of a black hole university-wise,” says Poulain, adding that discussions with a private university are already under way.
“We want the town of Sherborne to be known for its progressive AI and entrepreneurship”
Sherborne is also embedding AI in the classroom. It has developed its own platform, Sherbot, and is piloting Turing Tutor, an AI-powered learning platform that, like Caterham’s RileyBot, gives pupils access to an individual AI tutor. “It’s ambitious and very exciting,” says Poulain. Not only will Turing Tutor plug individual learning gaps, but it will also aid neuro-diverse students “who can soar in some areas and may struggle with others”, and give teachers a richer picture of each child’s progress by bringing together assessment and performance data.
“Ultimately we want the town of Sherborne to be known for its progressive AI and entrepreneurship,” says Poulain. “That means supporting not just our pupils, but parents, local businesses and the wider community. We’ve got a lot of young families in Sherborne but a lot of people wearing red cords do like to come down and retire here, and we want them to be good with AI as well.”
Being good with AI is what gets Ed Halliday out of bed every morning. Halliday is a lecturer in AI and entrepreneurship at the University of Bristol’s Centre for Innovation and Entrepreneurship, and co-founder of Future Faculty, an education start-up that equips young people with the skills and mindsets to thrive in future careers. “Naturally, AI has become an increasingly important part of that work,” he says. The company now runs AI boot camps for students and graduates, helping them become AI-native operators and gain an edge in a competitive jobs market. More than 4,000 students have taken part to date.
“Our hypothesis is that if you learn how to harness AI in the right way, you can say to an employer: ‘I’ve augmented my existing skills with AI to do more, faster, and achieve greater impact,’” says Halliday. “It’s about giving students a competitive advantage in a pretty complex labour market.”
Future Faculty has worked with the LSE, King’s Business School, UCL and other universities. Students aged 18-25 learn to build apps, prototype ideas and solve real problems set by employers, from designing products and marketing campaigns to creating automated business systems. “It’s a real-life sandbox,” says Halliday. “It’s about bringing the outside world into education.”
But AI is never the end goal. It is the process that matters, says Halliday. “Understanding a problem, interrogating it, testing ideas and refining solutions – that’s the cornerstone of my teaching.”
“It’s about giving students a competitive advantage in a pretty complex labour market”
“The fact we’re teaching AI is incidental because really we’re teaching critical thinking,” he adds. Students need to understand how AI models are built and “how they’re incentivised to produce information” so they can become “pilots rather than passengers”.
Halliday uses LLM ‘hallucinations’ as an example. To a lay person like you or me, hallucinations (when an LLM generates false, misleading or completely fabricated information, but presents it confidently as a factual truth) are a mistake, a fault in the system. Not so, says Halliday. It’s because these LLMs were built using something called “deep reinforcement learning”, which basically rewards a correct move but sanctions a wrong move, leading the LLM to logically guess when it doesn’t know the answer. “It’s not a bug,” says Halliday. “It’s actually a design feature.”
Future Faculty also encourages students to retain the “friction of the cognitive process” in their work. Before turning to AI, they discuss ideas with peers and sketch them out on paper, reflecting consciously on when AI adds value and when they want to rely on their own creativity. “It’s about thinking: ‘Where do I want to retain my humanity?’”
Despite the disruption AI is bringing, Halliday remains optimistic. “We’re in for a messy three-to-five years and then we’ll come out of it with more clarity.”
While he accepts the recruitment process is currently “broken”, his advice to young people is to find a different way. Don’t apply to the job on LinkedIn that has already had 100 applications – do your research, find the companies in your area and approach them directly. Don’t ask for a job; rather, let them know “I can help you”, and then show them how. AI has its limits, he says, and employers continue to value “human perspectives and human judgement”.
“I think we’ve reached a tipping point when we’ve realised we’re better than technology sometimes makes us feel,” he says. “We aren’t satisfied with seeing sloppy AI generated videos or posts on social media, because we enjoy things that are quite arduous, hard work, like sketching, using pen and paper or spending time discussing things. We’re going to run out of patience with copying and pasting AI answers into our homework, because it feels like giving up.”
David Flower, Warden Elect at St Edward’s Oxford, concurs. The school – affectionately known as Teddies – has taken a measured approach to AI education.
“The way we have approached it is that, if you’re interested in AI, computer science or STEM, you need the basic qualifications first,” says Flower. “The main problem with generative AI is that it bypasses the educational process, so there is a risk that pupils have gaps in their knowledge and become infantilised by the AI because they have become dependent on it as a shortcut.”
“The main problem with generative AI is that it bypasses the educational process”
Nor does he see much value in teaching prompt engineering. By the time today’s pupils leave school, LLMs may no longer require written prompts. “Prompt engineering is a bit of a gimmick,” he says. “It’s a bit like telling people how to use Google to search... Writing a good prompt is dependent on your scheme of knowledge.”
For Flower, a knowledge-rich curriculum is the foundation of effective AI use. Teddies has taken a slow-and-steady approach to making its curriculum and assessment “AI ready and relevant”. “We’ve been very careful not to throw the baby out with the bath water,” he says. “Education is a long-term project, and you can’t just be blown by the winds of changes in this or that direction.”
That means designing assessments that cannot easily be shortcutted by AI. Following an audit of its internally assessed Pathways and Perspectives course, for example, the school has shifted towards verbal presentations, handwritten work and sketchbooks, “things that we know are the work of the pupil”. It has also ended unsupervised typed assessments.
Teddies is not ignoring AI, however. From this term, sixth formers will work with Global Shapers, an AI literacy programme developed by graduates. “Our main aim is to make pupils critical users of AI,” says Flower.
He believes there is currently too much hype around AI. Young people are concerned about jobs, an education “outsourced to AI” and the world they are growing up in “being radically different from the world that existed 20, 30 years ago”. What pupils really want, says Flower, is some sense of where this is all heading, “which is, of course, the most difficult thing to do”.
But he argues that older generations and the media risk fanning the flames of AI anxiety. “There is a very lazy narrative that AI is taking young people’s jobs,” he says, pointing out that wider social changes, including the rise of remote working, have also reshaped entry-level employment.
“It’s very easy to view everything through the prism of AI, but society is much more complex than that.”
This article first appeared in The Week’s Independent Schools Guide Autumn/Winter 2026, edited by Amanda Constance.