Is Learning to Code Still Worth it in the Age of AI?

Is It Still Worth Learning to Code in the Age of AI?

85,096. That's how many students sat GCSE Computing this summer across England, Wales and Northern Ireland, which is down 7.1% from 91,619 in 2025 [1]. That's now the second year in a row that entries have dropped, with Ofqual's provisional figures for England showing a peak of 93,980 entries in 2024 that have fallen to 83,330 this summer [2].

One popular explanation for this is advancements in AI - if a chatbot can produce a working program in seconds, why would you spend two years learning to write one yourself? That's a pretty reasonable question to have today.

I have a degree in Computer Science, and for work, I train and evaluate cutting-edge AI models. My answer is yes; it's absolutely still worth it. The reasons have shifted over the last few years, though, and they're worth understanding before your child fills in their options form.

Why Are Fewer Students Choosing Computing?

Ofqual's chief regulator, Sir Ian Bauckham, offered a fairly convincing explanation when results came out. Many of this summer's students chose their options in 2024, when Generative AI was still relatively new, and he suggested that a coding-heavy course may have looked less necessary to students expecting AI to do the coding for them [1]. The managing director of Cambridge OCR also raised AI as a possible factor, whilst being clear that it was speculation [1].

The same regulator also said it was "difficult to think of subjects that could be more important" at the moment [1].

I agree with him. To some extent, I think the drop says more about how the subject is perceived than about what it actually involves.

AI Is the Future of Coding, So Someone Needs to Steer It

My own view is that AI absolutely is the future of coding; it'll write code faster and better than most of us, and on plenty of well-defined tasks it already does. The bigger shift is agentic AI. Asking a regular chatbot like ChatGPT for help is a bit like asking a friend for a recipe. Agentic AI is more like hiring a chef who plans the menu, does the shopping, cooks and tastes as they go. It breaks a task into steps and works through several parts of a project at once, testing and fixing its own work along the way.

Where it still falls short, though, is the bigger picture; AI can keep track of lots of moving parts just fine, but it struggles to hold a clear, creative vision for a whole system and follow it through every decision the way a person can. That part is still the human's job. Because of this, the job of a programmer has shifted from typing every line towards directing AI and checking and modifying what it produces.

You can't do that well without understanding code yourself, though, so knowing how to code is what lets you tell AI exactly what you want and spot when it hasn't delivered, which is where the next example comes in.

The Bug That Looks Completely Fine

This short Python function is meant to find the lowest temperature in a week of readings. Starting the lowest value at 0 is a classic mistake I see in lessons, and an AI tool could just as easily produce it if the request is vague. 

def find_lowest(temperatures): 
    lowest = 0 
    for temp in temperatures: 
        if temp < lowest: 
            lowest = temp 
    return lowest 

Test it on a January week in Newcastle, [-2, 1, -4, 3], and it correctly returns -4. It runs without errors, it's neatly laid out, and it passes the obvious test. Now try a week in July, [14, 17, 12, 19].

Tracing It Line by Line

The function returns 0, a temperature that never happened that week. To spot this problem, look to line 2. Starting at 0 assumes at least one reading will be below freezing, which, even in North East England, isn't very likely in July! Starting from the first reading - lowest = temperatures[0] - fixes it. A careful programmer would then ask what should happen if the list is empty.

Tracing code by hand like this is taught directly in GCSE Computer Science [5], and that's one of the skills you'll need to check an AI's output before you trust it. For extra practice with coding concepts, consider focused help in GCSE Programming.

Even Experienced Developers Are Leaning on It

In early 2025, the research group METR ran a randomised controlled trial and found that experienced open-source developers took 19% longer to complete real tasks when AI tools were allowed, despite forecasting beforehand that AI would cut their time by 24% [3]. A year is a long time in AI, though. By February 2026, METR had labelled those results out of date and said developers were now likely being sped up by AI tools [3][4]. So many developers refused to work without AI that METR is now redesigning the study, and some were running several AI agents at once [4].

What stands out to me most, though, is who those developers are; they're experienced programmers with a median of 10 years behind them [4], and in my view, that's exactly why they get so much out of AI. They know what good code looks like, so they can direct it, check it, and step in when it goes wrong.

GCSE Computer Science Is More Than Coding

Although programming is central to the course, it's actually just one part of it. OCR's GCSE, for example, has two exams, one on computer systems and one on computational thinking, algorithms and programming [5]. The first covers how processors, memory, networks and cyber security work, along with the ethical and legal side of technology [5]. BCS, the Chartered Institute for IT, made a similar point this summer, arguing that computing education is about computational thinking, problem solving and understanding digital systems, with the programming language only one piece [1].

The Numbers at a Glance

• 85,096 GCSE Computing entries in 2026 across England, Wales and Northern Ireland, down 7.1% on 2025 [1]

• 29.9% of students in England achieved a grade 7 or above, up from 28.3% in 2024 [6]

• 70.6% achieved a grade 4 or above, up from 68.3% in 2024 [6]

• 36.0% of girls achieved a 7 or above, compared with 28.1% of boys, even though girls made up only 23.0% of entries [6]

It's also fair to say this is quite a demanding GCSE. The Curriculum and Assessment Review found that, on average, both boys and girls achieve lower grades in Computer Science than in their other GCSE subjects [7], so it's worth considering whether you believe you'll enjoy the subject instead of assuming it's an easy GCSE.

The Course Is Changing, But Not for This Year's Year 9s

Last November, the government committed to replacing the Computer Science GCSE with a broader Computing GCSE, and also to exploring a new Level 3 qualification in data science and AI [8]. The new GCSE will keep core principles like programming and algorithms, while the wider computing curriculum adds AI literacy, working with data and bias in technology [8], which is a much-needed update given the pace AI is currently moving at.

This timing is important for families choosing options now. Reformed GCSEs are due in two phases, with first teaching in 2029 and 2030 [9], and none will be taught before September 2029 [8], so a student starting GCSEs in September 2027 will sit the current course. It's also worth noticing which direction the reform points, as the government's stated aim is to encourage more pupils to consider computing qualifications [8].

Is It the Right Choice for Your Child?

In my experience, the students who enjoy GCSE Computer Science most tend to have a few things in common.

• They like puzzles, logic problems or working out why something isn't doing what it should.

• They're reasonably comfortable with maths; by no means do they need to be top set, but algebra and number work do very often come up — and targeted GCSE Maths support can help.

• They're persistent, because a lot of programming is fixing things that didn't work the first time.

• They're curious about how the technology they use every day actually works.

Previous coding experience helps, but it definitely isn't a requirement.

If Your Child Already Uses AI

That's perfectly OK, and learning to use it well is part of the point; it's not like it's going away anytime soon. When revising for written programming exams, I often encourage my tutees to use AI to explain a piece of code they don't understand, then close the chat and write it themselves from scratch. If they can't, they haven't understood it yet, which helps identify areas that need more revision. This doesn't mean they should be using it to look up answers to exam questions so they don't have to do the thinking themselves; it's about using AI selectively and responsibly to make the most out of their revision time.

The Short Answer

AI has changed what programming looks like day to day, but it hasn't removed the need to understand what a program is doing. In my own job, the skill I lean on most is noticing when something that looks right is wrong, and that's exactly what GCSE Computer Science starts building from Year 10.

If your child is weighing up their options, or has started the course and is finding it a bigger step than expected, you're very welcome to get in touch through my Sherpa profile. I'm always happy to talk it through.

Sources

1. Tes, "AI may repel students from taking GCSE computing, exam board boss warns", 21 August 2026. https://www.tes.com/magazine/news/general/ai-repel-students-taking-gcse-computing-ocr-boss-warns

2. Schools Week, "GCSE computing entries drop by 10,000 in two years", 9 June 2026. https://schoolsweek.co.uk/gcse-computing-entries-drop-by-10000-in-two-years/

3. METR, "Measuring the Impact of Early-2025 AI on Experienced Open-Source Developer Productivity", 10 July 2025. https://metr.org/blog/2025-07-10-early-2025-ai-experienced-os-dev-study/

4. METR, "We are Changing our Developer Productivity Experiment Design", 24 February 2026. https://metr.org/blog/2026-02-24-uplift-update/

5. OCR, GCSE (9-1) Computer Science J277 specification. https://www.ocr.org.uk/Images/558027-specification-gcse-computer-science-j277.pdf

6. National Centre for Computing Education, "GCSE exam results 2026, highlights for computer science", August 2026. https://teachcomputing.org/blog/gcse-results-2026

7. Department for Education, Curriculum and Assessment Review Final Report, November 2025. https://www.gov.uk/government/publications/curriculum-and-assessment-review-final-report

8. Department for Education, Government response to the Curriculum and Assessment Review, November 2025. https://www.gov.uk/government/publications/curriculum-and-assessment-review-final-report-government-response

9. AQA, "Curriculum and Assessment Review final report: our response and what it means for you", November 2025. https://www.aqa.org.uk/about-us/what-we-do/policy/our-response-to-the-curriculum-and-assessment-review