Why AI Is the Future of A-Level Maths Revision (And Why Old Sites Are Falling Behind)

The revision tools that defined a generation — and their ceiling

For the past two decades, A-Level Maths revision looked the same:

1. 1Download past papers from the exam board website
2. 2Watch YouTube videos when stuck
3. 3Read through textbook examples
4. 4Repeat

This works. Tens of thousands of students got A*s using exactly this approach. But it has a ceiling — and in 2025, that ceiling is increasingly visible.

The fundamental problem with static resources

Every past paper ever written has already been sat by hundreds of thousands of students. Mathswatch has the same videos it had in 2018. Physics & Maths Tutor's question sets were uploaded years ago.

Static resources create a specific problem: familiarity masquerading as understanding.

When you do the same 2019 Edexcel Paper 2 for the third time, you're not practising — you're remembering. Your brain starts pattern-matching on the layout, the question numbers, even the font. This is why students can score 85% on a paper they've seen before and 58% on a fresh one.

Real exam performance requires encountering questions you've never seen. Until recently, the only way to get those was to wait for new past papers — which appear once a year.

What AI changes

Infinity Stars uses Claude (Anthropic's frontier AI) to generate exam-style questions in real time. Not templates. Not shuffled numbers. Genuinely novel questions, constructed to match the exact style, depth, and mark allocation of your specific exam board.

Here's what that means in practice:

Infinite practice volume — You never run out of unseen questions. A student revising integration by substitution can generate 50 new questions in a session, each genuinely fresh.

Board-specific calibration — An Edexcel question has different phrasing, different command words, and different mark allocation expectations than an AQA question. Infinity Stars generates questions that feel indistinguishable from real exam papers for your board — not generic questions that happen to involve calculus.

Instant worked solutions — Every question comes with a full mark scheme and per-mark method guide. Not just the answer — the exact reasoning at each step, including what the examiner is looking for.

A* timing benchmarks — Static resources give you questions but no speed reference. Infinity Stars shows you, after every question, how your time compares to an A* student. This is data that has never existed before in revision.

Why established sites haven't kept up

Physics & Maths Tutor is a brilliant archive. It contains virtually every past paper, mark scheme, and topic question ever produced. That's genuinely valuable.

But it is an archive. It can't generate new content, it can't personalise to your exam board, it can't benchmark your timing, and it can't explain why each mark is awarded.

Mathswatch is video-based. Videos are passive. You watch someone else do the maths and feel like you're learning. You're not — you're observing. The only way to build mathematical ability is to do mathematics.

These sites were built for a world before AI. They haven't changed because changing would require rebuilding from scratch. Infinity Stars was built from scratch for this world.

The AI advantage in exam preparation

There's a concept in educational psychology called "desirable difficulty." Learning is maximised when material is hard enough to require genuine effort — but not so hard that failure becomes demoralising.

Static past papers can't calibrate to this. Too easy (a paper you've seen) or potentially too hard (a topic you haven't revised recently mixed into an unfamiliar format).

AI-generated questions can be dialled to exactly the right difficulty for where you are right now. Generate beginner questions when starting a topic, standard questions as you progress, challenging questions in the final weeks.

What this means for your grade

Students who use active, AI-powered retrieval practice outperform students who use passive, static resources. The research is clear on this — and the grade data from Infinity Stars students supports it.

The question isn't whether AI will transform A-Level revision. It already has. The question is whether you'll use it.