AI Summarisation Meets Revision Season: The Unlikely Return of the One-Page Cheat Sheet
Every May, the library fills up and the same debate resurfaces: what actually works for revision? This year the conversation has a new participant. Generative AI tools have been folded into nearly every step of student workflows — and yet the format they are increasingly used to produce is one of the oldest in the book: the dense, one-page cheat sheet.
The evidence never favoured re-reading
Cognitive science has been unambiguous on this for decades. Passive re-reading and highlighting are among the weakest revision strategies; what works is retrieval practice — forcing yourself to recall material — and the act of condensing notes, which requires deciding what matters. The classic studies on testing effects by Roediger and Karpicke, and the broader literature on "desirable difficulties," all point the same direction: effort during preparation is the feature, not the bug.
The humble cheat sheet sits in a sweet spot. Some courses permit a single A4 reference sheet in exams precisely because instructors noticed that students who make one often barely need it by exam day. Compressing a term's lectures onto one page is retrieval practice disguised as an arts-and-crafts project.
So where does AI fit?
The criticism writes itself: if an AI condenses the notes for you, haven't you skipped the useful struggle? Partly, yes — and students who outsource the whole job will learn what everyone who outsources studying learns. But the emerging usage pattern is more interesting than wholesale delegation.
Students are using AI to produce a first draft of the condensed sheet — extracting the formulas, definitions, dates, and edge cases from a 300-page PDF of lecture slides — and then doing the genuinely valuable work on top: cutting what they already know, reorganising by weakness, rewriting in their own shorthand. The AI handles the secretarial layer; the student keeps the cognitive layer.
The tooling has caught up with the pattern. Free utilities like ChatSlide's AI cheatsheet maker take a lecture PDF and return a sectioned, one-page draft — terse reference lines under scannable headings rather than flowing summary prose — which is exactly the right starting material for the editing pass that does the teaching. (The same site's document tools solve the adjacent end-of-term problem: a dissertation PDF that exceeds the submission portal's upload limit can be shrunk with a PDF compressor without re-exporting from Word.)
A pattern worth noticing
There is a broader lesson here about how AI is actually settling into study practice, as opposed to how it was predicted to. The doom scenario — students who never engage with material at all — exists, but so does its inverse: AI stripping away the low-value mechanical work that previously consumed revision time. Typing up notes, reformatting, hunting through slide decks for that one formula: none of that was ever the learning. The condensing, connecting, and self-testing was.
The one-page cheat sheet survives because it forces exactly that. The tools changed; the pedagogy, stubbornly, did not. Students heading into this exam season could do worse than the old advice with a new first step: generate the draft, then make it yours — and let the making be the revision.