Admissions on predicted A-level grades look foolish in AI era

Today’s publication of A-level results marks the start of a conversation familiar to UK campuses in mid-August. How will clearing perform? Which universities are recruiting most aggressively? How far will entry tariffs quietly soften? Which subjects will fill first? How many applicants will change their minds on results day?

It’s become an annual ritual. We scrutinise recruitment strategies, analyse the numbers and debate whether the admissions system worked as intended. Yet for all the discussion about how clearing operates, we rarely step back and ask a more fundamental question: why does the UK need clearing at all?

This isn’t an argument against either clearing or Ucas. Quite the opposite. The creation of Ucca (the University Central Council on Admissions), the predecessor to Ucas, transformed university admissions. Before the early 1960s, university admissions were fragmented, with students making multiple direct applications and institutions competing for applicants. A coordinated national admissions system was a huge step forward and remains one of the defining features of UK higher education.

The question, then, is not whether Ucas works. It plainly does. Rather, it’s whether one of its defining design choices still makes sense: asking students to apply before they know their final examination results.

The design made perfect sense at the time. Universities processed paper applications, admissions decisions took months and institutions needed an incoming cohort long before the academic year began. Conditional offers based on predicted grades were an elegant response to those practical constraints.

Sixty years on, however, the admissions landscape looks very different from the one that Ucca was designed to serve.

The Sutton Trust argued as far back as 2020 that the UK was the only major higher education system to base admissions primarily on predicted rather than achieved grades. Since then, the evidence has been remarkably consistent. A 2024 peer-reviewed study found that fewer than one in three predicted grades exactly matched the grade achieved, while Ucas’ analysis of the 2025 admissions cycle highlighted a widening gap between predicted and achieved grades. About half of UK 18-year-olds were predicted AAA or above, yet only 26 per cent of accepted applicants ultimately achieved those grades.

None of this is a criticism of teachers. Predicting future performance will always involve uncertainty. But if the evidence continues to point in the same direction, perhaps the question is no longer whether predicted grades can be made more accurate. Perhaps it’s whether university applications should depend on predicting future attainment at all.

Internationally, most comparable countries have reached a different conclusion. Students in Germany apply with their achieved Abitur results. In Australia, university entry is largely determined after students receive their final ATAR. In China, applicants receive their gaokao scores before expressing their university preferences. These systems differ in many respects, but they share one important characteristic: students know what they have achieved before deciding where to apply.

That doesn’t automatically make them better. The UK’s approach has significant advantages.

Universities don’t gain certainty; they gain time. An early view of likely student numbers allows institutions to plan staffing, teaching space, accommodation and student support. Applicants benefit too, receiving reassurance months before results day. There is also a widening participation argument: for many applicants, particularly those who are the first in their family to consider university, an early offer provides confidence, validation and something tangible to work towards during their final year at school.

These are genuine strengths of the current system. But they are balanced against an equally important consideration: students are still making one of the biggest decisions of their lives based on what they are predicted to achieve rather than what they have actually achieved. The more I’ve thought about this, the more I’ve found myself wondering whether we’ve simply accepted this as the way admissions works.

Personal statements have been replaced by structured application questions, contextual admissions have become increasingly data-driven, and artificial intelligence has fundamentally altered the value of written application materials. At the same time, achieved qualifications have become, if anything, an even more important objective measure of academic attainment.

Universities are also operating in a different recruitment environment. A lengthy admissions cycle has undoubtedly helped institutions plan their intake, recruit internationally and support applicants through visa processing, accommodation and enrolment. But as government policy seeks to reduce reliance on international fee income, it is worth asking whether the balance of advantages is beginning to shift.

None of this is an argument for abolishing clearing, nor is it a call to dismantle Ucas. A coordinated national admissions system remains one of the great strengths of UK higher education, and every admissions model involves trade-offs. Different countries have arrived at different solutions because they operate within different educational, political and cultural contexts.

The most common objection to applications based on confirmed grades is that there isn’t enough time between A-level results day and the start of the academic year. That’s a perfectly reasonable concern, but it also assumes every other part of the system is fixed. Students sit examinations in May and June yet receive their results in mid-August, while universities begin teaching a few weeks later because that’s how the academic calendar has evolved. If we were redesigning admissions around confirmed achievement, should those assumptions also be open to challenge? A later academic year, earlier results as assessment technology develops, or a combination of both may all be possible. The point isn’t to advocate one solution, but to recognise that redesigning admissions may require us to rethink the wider system too.

What strikes me, however, is that we’ve become very good at debating the mechanics of admissions without revisiting one of its underlying assumptions.

For years we’ve debated how to improve clearing. Today we’re debating artificial intelligence. They’re important conversations, but perhaps they’re symptoms rather than the underlying issue.

If we were designing a national admissions system from scratch in 2026, knowing everything we now know about technology, assessment, applicant behaviour and admissions practice around the world, would we really ask students to apply on the basis of predicted grades rather than confirmed achievement?

I’m not convinced we would.

Gianina Harvey-Brewin is dean of continuing education at the University of Roehampton and a board director of the Institute of Student Employers.