GCSE results and AI are telling two very different stories about UK education right now. One story is essentially flat, the other is accelerating fast. In 2026, AI tools capable of explaining, drafting and solving almost anything on the syllabus became more powerful and more widely used than ever. Over the same period, national GCSE pass rates barely moved, and in some subjects, they slipped.

That is a genuinely odd combination. If AI really does help students learn, shouldn’t results be improving?

What The 2026 GCSE Results Actually Show

This summer, 320,505 students in 2,835 schools and colleges received A-Level results, with 835,950 individual A-Level results issued in total. For GCSEs, 1,047,750 students across 6,040 schools and colleges received results, with 5,720,045 individual GCSE results issued. The difference is because most students receive results for several subjects.

What do the 2026 GCSE results actually show? The honest answer is: it is more nuanced than a straightforward decline, and it is worth being precise about it. The national proportion of entries achieving a grade 4 or above dipped only marginally, from 67.4% in 2025 to 67.3% in 2026. Top grades moved in the opposite direction: 22.0% of entries achieved at least a grade 7 or an A, slightly up on the year before.

Where the picture gets more concerning is Maths. Analysis of Ofqual’s figures suggests 178,270 16-year-olds in England missed a grade 4 in maths this year, 2,720 more than in 2025. English told a slightly different story, with 180,570 16-year-olds missing a grade 4, actually a small improvement of 2,900 fewer than the previous year.

So it is not accurate to say results are “getting worse” across the board. It is closer to a stall, with maths showing signs of genuine strain while English holds broadly steady.

The Resit Surge: The Clearest Trend In The Data

The part of this story that is unambiguous is what is happening to resits. Analysis of Ofqual’s figures suggests more than 350,000 current 16-year-olds will need to resit English or Maths next year. The resit system itself is under growing pressure: post-16 entries for English rose 6.5% on the year before, but the pass rate fell 1.4 percentage points; for Maths, entries rose 5.8% while the pass rate fell 1.7 percentage points on 2025.

More students are having to retake these GCSEs, while success rates fall with each attempt. That creates a growing problem, just as AI tools are supposedly making learning support more accessible than ever.

Why Hasn’t AI Moved The Needle Yet?

There is no single proven explanation, but a few honest possibilities are worth naming.

The curriculum has not caught up yet. The government has accepted recommendations to add AI, data science and stronger digital literacy to the curriculum. However, curriculum reform moves slowly. Evidence given to a parliamentary committee warned that, allowing for writing new specifications, developing qualifications and running through a full two-year GCSE and two-year A-Level cycle, meaningful change for students leaving school could realistically be seven years away. In other words, the AI-literacy benefits of curriculum reform are still mostly ahead of us, not behind us.

Access to AI is not the same as knowing how to use it well. As we covered in our piece on AI and homework, the students who get genuine value from AI are the ones who already understand a subject well enough to question and check what it gives them. For a student without that foundation, AI can produce something that looks like progress without building any of the understanding an exam actually tests.

Schools are, rightly, restricting AI where it matters most. Ofqual has been explicit that AI must not be the sole marker of any assessment component, including coursework, and has written directly to exam boards over concerns about AI misuse in non-exam assessment, particularly in extended-writing subjects like History and English. That means any theoretical “AI boost” is being deliberately limited in exactly the assessments where it could most easily be misused. This is the right call for exam integrity, but it also means AI is not (and should not be) quietly propping up grades.

What This Means For GCSE and A-Level Students Right Now

If the resit numbers show anything, it is that gaps in Maths and English are best caught early, not patched later. A few practical takeaways:

• Do not wait for the resit stage. The jump in resit entries suggests many students are only discovering a foundational gap after their first GCSE attempt. Targeted support during Years 10 and 11, well before results day, is far more effective than trying to close the same gap post-16.
• Use AI to test understanding, not replace it. The same principle from our AI-and-homework piece applies directly here: AI is genuinely useful for generating practice questions and checking your own answers, and much less useful, arguably counterproductive, as a substitute for the process of learning the material in the first place.
• Get a second, human perspective on where the gaps actually exist. A tutor can diagnose why a topic is not landing in a way that generic AI explanations cannot, which matters most in exactly the subjects, Maths and English, where this year’s data shows the most strain.

The Apollo Scholars Conclusion

This year’s results are a useful, real-world instance of the exact point we made in our AI-and-homework article: access to better tools does not automatically produce better outcomes if the underlying understanding still has to be built the slow way. AI is more capable than ever, and results are still barely moving, because grades were never really about access to information. They are about whether a student has actually built the knowledge to use that information well.

If your child is heading toward a resit, or you would rather catch a gap in Year 10 than fix it in Year 12, that is exactly what our tutoring is built to achieve. For revision help in the meantime, our Store has additional resources, and our podcast, Mission Control, breaks down tricky topics in plain, exam-relevant terms.