In a Cambridge classroom, 10-year-old Joseph trained his AI model to distinguish between a picture of an apple and a picture of a smile.
“The AI gets a lot of things wrong because it mistook fruit for a face,” he said. He began to retrain it and, in one fell swoop, he got it back on track – intuitively understanding the intrinsic nature of artificial intelligence and machine learning, the way few adults do.
His friends from the St Paul’s C of E Primary School coding club attempted to create their own AI with similar ingenuity. Just as people born in the early 19th century never knew a world without unmanned flight, and Generation Z has always lived with social media, Joseph and his friends are AI natives.
Here, on a December morning, some of them were being taught the principles and practicalities of a potentially world-changing technology that experts fear could infect large numbers of people and leave them powerless.
Philip Colligan, chief executive of the digital education charity Raspberry Pi Foundation, has warned of a “huge divide” in society between those who understand how AI work and those who are able to control them – challenging their growing role in automating decisions in areas including housing, welfare, health, criminal justice and finance. On the other hand, there may also be a cadre of AI illiterates who risk social disempowerment.
Colligan, a leading expert on technology and its social impacts, told the Guardian that AI literacy must become a universal part of education, on the same level as reading and writing, to prevent social divides from opening up.
“There’s a world where you get a huge divide between kids who understand, who have the basic knowledge and are therefore able to assert themselves, and those who don’t,” said Colligan, whose charity is affiliated with the £600m British low-cost tech hardware startup of the same name. “And that can be very dangerous indeed.”
His warning was supported by computer researcher Simon Peyton Jones, who led the creation of a schools national curriculum for computing in 2014, before the AI boom. He called for a new digital literacy qualification for all schoolchildren that would ensure they know how to use AI in significant ways.
“If it’s just a black box, (its actions) seem like magic,” he said. “It’s very unforgiving if you don’t know anything about how magic is working. I’m very concerned about students leaving school without agency in the world.”
His comments come amid a decline in the number of children studying computing, with 2025 entries for GCSEs in the subject across the UK. Today, three times as many people take history and almost twice as many take biology, chemistry, and physics. Plus, the use of AI systems is on the rise across the country — up 78% in the year to September, according to voting By Ipsos.
Part of the perception that learning computing skills is becoming redundant comes from some big AI companies, who argue that their systems are going to automate coding. Anthropic’s chief executive, Dario Amodei, said in October that 90% of its coding was automated using its cloud AI models. Meanwhile, 2025 was the year that “vibe coding” became a common phrase – connoting the idea that AI would allow humans to create software using natural language instructions rather than specialized code.
Political leaders such as Keir Starmer have also suggested that coding is becoming redundant. As Leader of the Opposition in 2023, he said: “The old way – learning old IT on 20-year-old computers – doesn’t work. But neither does the new fashion, that every kid should be a coder, when artificial intelligence will blow up that future.” This has led to the idea that understanding the inner workings of computers may become less relevant in the future.
“I think they’re just exaggerating the benefits,” said Colligan, whose charity works in schools in dozens of countries.
“The message is leaking out that kids don’t need to learn all this anymore and that’s not only flawed, but dangerous. We’re already talking to teachers in many schools, not just in the UK, but around the world, saying: ‘We can leave computer science now, right?’ This is a problem.”
He added, “We’re all moving into a world where more and more of the decisions we face every day will be made by automated systems. Right now it’s what movie should I see next or what song should I listen to? Very soon it’s going to be finance decisions, health care decisions, criminal justice decisions. You can’t advocate for your rights if you don’t understand how these decisions are being made by automated systems. You can’t challenge what’s presented to you.” You cannot critically evaluate what is being done.”
In December, former Deputy Prime Minister Nick Clegg, now an AI investor, predicted that “we will move from staring at the Internet, to living in the Internet”.
Colligan said: “My concern is that there will be a gap between kids based on their socio-economic background. Some kids who go to great schools, that are able to teach all this, will be in a much stronger position as citizens, whether they’re using technology for their work or not. Kids who are in communities where they don’t have access to[AI literacy teaching]will be on the receiving end of a whole load of passively automated decisions.”
In the coding club, children aged seven to 10 are taught how AI works. The lessons were clearly having an impact on Joseph. He said he thinks AI “will probably be good, but if too many people believe in it when it is wrong then it will have a bad effect on them”.
They had no interest in letting AI do the coding for the video games they planned to create. “It may be different than what you want,” he said. “It can also go wrong and you have to know how to solve it… I would like to be in charge of AI. If AI is in charge of us, we won’t really be able to control what we are doing and that would be bad.”
