A dangerous myth is gaining traction with teenagers picking their university subjects, which I’m sure has also spread to many parents: there’s no point studying computing any more, because Generative AI can do it all now.
On results day, figures from UCAS showed that the numbers of students accepting places on Computing degrees at UK universities had fallen for the second year running; (by seven per cent for UK 18 year olds).
While we also know that people taking dedicated degrees in AI have been rising, these represent a very small minority of all computer science graduates.
My professional body, BCS The Chartered Institute for IT, is asking government, employers and my sector – education – to take this trend very seriously.
What a computing qualification really means
Why? Not just because of the risk of expecting whole areas of critical infrastructure to run safely on vibe-coded software, without expert technologists who understand what it takes to maintain it
The reason is that it’s built on a misunderstanding of what a computing qualification has ever been for.
A degree in computer science was never about learning a particular programming language, framework, or paradigm, any more than chemistry has ever been about mixing things in test tubes. The subject is and always has been about understanding the world around us well enough to make new things that work, and improve the things that don’t. In computing, that understanding has a name: computational thinking, the ability to take a messy real-world problem and work out how to make a machine solve it. Languages and frameworks come and go. That skill does not.
I suspect the myth comes from a reasonable-looking but false analogy: you can type a sentence into an image model and get a passable picture for a funny birthday card, so it feels as though you ought to be able to type a sentence or two into some other AI chat thing and get a working system of any complexity out of the other end. You cannot. Building, debugging, securing and maintaining real software is nothing like ordering a cartoon.
Evolving at pace
Generative AI is a genuinely powerful addition to the toolbox; it can help you think, save you time, point you at problems you would have missed, but it is a tool alongside the others, not a substitute for knowing what you are doing.
There is something particular about this technology that makes the panic understandable though. The past fifty years or so have not been short of disruptive technologies, and every one of them left computing a solid career. But the earlier ones were at least explainable. The computer: a machine that follows simple instructions mindlessly, but very fast. The internet: a way for computers to talk to one another. The web: a way to publish and link content. Each explanation fits neatly into a short sentence. We also saw these technologies evolve from their most basic form to what we know now. Generative AI, on the other hand, seemed to arrive out of nowhere in a startlingly mature form.
It keeps evolving at pace, and it is annoyingly and stubbornly resistant to explanation. I have spent more than a year trying, with colleagues who know far more than I do, to find a way to describe how it does what it does that is both true enough and clear enough to be useful to a general audience. I have failed. When even the practitioners struggle to explain the magic, it is no wonder the creative world is anxious about what it means for them, and no wonder computing is asking the same question of itself.
Lessons from a history of disruption
We have been here before though, even if in a smaller way. What we ask computers to do has changed beyond recognition over the past few decades. COBOL code to batch-process a payroll was miraculous in its day, but is orders of magnitude simpler than the infrastructure behind a single modern website. No sensible person would build all of that from scratch now.
As the work grew more complex and computers became more powerful, we kept finding ways to factor out and standardise the boring, repetitive parts, so we could concentrate on the creative, research-driven, entrepreneurial and business-critical ones. It would be unfair to generative AI to call it just the latest of these – it is more than that – but there is a family resemblance. It will not have the next world-changing idea for you. It will help you build that idea, if and only if you know what you are doing. It will help you critique it and debug it. And when the simple prompt “fix my code” stops working, it is the person who actually understands digital systems who becomes irreplaceable.
New paradigm, old skills
Which is the whole point. The most valuable thing a computing education gives you is not fluency in language X or framework Y. It is the ability to reason about how real-world concepts map onto the digital world, to think computationally, and AI raises the value of that skill rather than lowering it. The labour market is already saying so: demand for people who genuinely understand systems is climbing even as routine coding is automated away.
So the honest message to that sixteen-year-old is the opposite of the myth.
More and more people now need to understand enough of this discipline to use the new tools well, and to recognise when they are being led astray. And because AI and digitisation are reshaping every corner and crevice of society, we need more people, and a wider more varied range of them, going deeper into it, not fewer.
Writing code was always the easy part.
