AI tools are everywhere in education now. Most students I talk to are using at least two or three of them regularly, whether that’s for writing essays, looking things up, or organising notes they should probably have organised weeks ago. The problem is that a lot of these tools are being used in ways that make studying less effective, not more.
There’s a distinction worth making, and I don’t think enough students have thought about it. Some AI tools do the thinking for you. Others force you to think. The ones that help you learn tend to be the second kind.
## The passive trap
Here’s what I see happening a lot, particularly with GCSE and A-level students gearing up for exams. A student has a textbook chapter to revise. They paste it into ChatGPT, ask for a summary, read the summary, and feel like they’ve studied. They haven’t. They’ve let a machine do the thinking, and the thinking was the whole point.
This isn’t a moral failing. It’s a reasonable response to the tools available. But it runs directly against what cognitive science tells us about how memory works. Dunlosky and colleagues reviewed ten common study techniques in a 2013 paper and found that the two most effective were practice testing and distributed practice. Rereading and highlighting, the methods most students default to, were rated low effectiveness.
The issue with using AI to summarise your notes is that it replaces the mental effort that actually builds memory. You end up with a tidy summary you didn’t have to think about. That might be fine for getting a rough overview before you start, but it’s not the same as revision.
## What the research says about testing yourself
The reason self-testing works so well comes down to something called the testing effect. Roediger and Karpicke published a study in 2006 that showed students who took practice tests after reading a passage remembered significantly more a week later than those who simply re-read the material. Pulling information out of your memory, even when you get it wrong, strengthens the memory trace in ways that going over notes again does not.
This has been replicated many times since. A 2024 systematic review of active recall strategies found that practice testing and self-quizzing were reliably linked to better academic outcomes in young adults. At this point, retrieval practice is one of the most well-supported findings in educational psychology.
So when you’re picking AI tools for revision, it’s worth asking: is this tool making me work to remember things, or is it remembering them for me?
## Tools that actually help
With that in mind, here are the tools I’d actually recommend.
Before you can test yourself on anything, you need your notes in one place. Notion works well for this because it handles different formats: lecture recordings, PDFs, handwritten notes you’ve photographed. The AI search features are handy for finding things later, but I’d avoid using Notion to generate summaries you haven’t written yourself. Organisation is the boring part of studying, but skipping it leaves gaps.
The self-testing part is where AI is most useful. Writing practice questions from your own notes by hand works, but it takes ages, especially when you’ve got four or five subjects on the go. A quiz generator can take your notes and produce multiple choice or short answer questions in seconds. Quizgecko is one tool that does this. You paste in your material, pick a format, and start testing yourself straight away. The questions won’t always be perfect, but they force you to retrieve what you’ve studied rather than just stare at it again.
For spaced repetition, Anki is hard to beat. It’s been around for years. The idea is that you review material at gradually increasing intervals, so you come back to each card just as you’re about to forget it. It’s particularly good for factual recall. A-level biology terms, history dates, that sort of thing. The interface is ugly, I won’t pretend otherwise. But the scheduling algorithm works and it’s free.
Grammarly is worth mentioning with a caveat. It’s useful for catching grammar mistakes and awkward phrasing in your own writing. It becomes a problem when students use it to rewrite entire paragraphs they didn’t compose themselves. There’s a difference between getting feedback on your writing and having software write for you. Be honest with yourself about which one you’re doing.
For research, Google Scholar is still the best starting point when you need peer-reviewed papers for essays and coursework. Perplexity is newer and handy for getting an overview of a topic with citations, though I’d always check the sources it gives you. Neither writes your essay for you. They help you find material, and you do the reading.
## Building a weekly revision routine
Tools on their own don’t do much. How you fit them together matters more than which ones you pick.
A rough weekly pattern that works: spend the first half of the week studying new material and taking notes in your own words. Mid-week, generate practice questions from those notes and test yourself without looking at the source. Pay attention to what you got wrong. Spend a session going back to those weak areas specifically, then test again. At the weekend, keep up with your Anki reviews and add new cards. Twenty minutes is enough if you’re consistent about it.
This doesn’t need to be rigid. The point is that self-testing happens throughout the week, not just the night before an exam when you’re panicking.
## A note on academic integrity
UK universities and exam boards are updating their policies around AI use, and the rules vary quite a bit depending on your institution. The Russell Group universities generally take the position that AI can be used for personal study and learning, but submitting AI-generated content as your own work in assessed coursework is academic misconduct.
The same principle applies at GCSE and A-level. Using a tool to quiz yourself on your notes is studying. Pasting an essay question into ChatGPT and submitting what comes back is cheating. The line between the two is usually obvious, even if it’s tempting to pretend otherwise. If your school or university has published specific guidance on AI use, read it. And when in doubt, ask your teacher or lecturer.
## The right tools don’t replace the work
The way I think about AI study tools is that they’re multipliers. If you’re already testing yourself and spacing your revision, they can speed the process up and take some of the admin out of it. If you’re using them to skip the hard parts of studying, you’ll probably get less out of your revision than someone with a pen and a blank sheet of paper.
Learning is supposed to feel difficult sometimes. That moment where you’re staring at a blank page trying to remember something you read yesterday, and it’s not coming, that’s frustrating. But it’s also where the actual learning happens. I’d rather students picked tools that lean into that difficulty than tools that paper over it.