{"course":{"slug":"ai-for-teachers","title":"AI for Teachers","url":"https://addaly.com/learn/ai-for-teachers","level":"beginner","track":"work"},"lesson":{"slug":"revision-flashcards-and-spaced-practice","title":"Revision: flashcards and spaced practice","url":"https://addaly.com/learn/ai-for-teachers/revision-flashcards-and-spaced-practice","position":24,"of":57,"minutes":9,"answer":"The evidence for retrieval practice and spacing is strong and the model can generate a term's flashcards and a spaced schedule in minutes; the failure is cards that test recognition rather than recall.","keyIdea":"The evidence for retrieval practice and spacing is strong and the model can generate a term's flashcards and a spaced schedule in minutes; the failure is cards that test recognition rather than recall, and the fix is a card format that forces the student to produce the answer."},"body":{"format":"markdown","text":"## Two findings that survive every replication\n\nTwo results from the research on memory hold up wherever they are tested. Retrieval practice, trying to produce an answer from memory, builds lasting memory far better than rereading, even when the rereading feels more productive. Spacing, the same material returned to over days and weeks, beats the same total time spent in one block. Neither needs technology. Both are made much easier by it.\n\nThe model's role is to generate the material for retrieval at volume. Your role is to make sure the material actually forces retrieval, which is the thing it gets wrong.\n\n> Figure: Why the effortful version wins later. Line chart of % of items recalled in a test against Days since the lesson: Read the notes three times (1, 79) (7, 52) (14, 40) (28, 29); Answered from memory, spaced (1, 74) (7, 67) (14, 62) (28, 56). Both groups spent the same total time. Rereading feels more productive on the day and is measured lower a month out; producing the answer from memory, spread over days, holds. The shape is the well-replicated one; the exact numbers depend on the material. This is why a card with four options on the back teaches recognition and not recall.\n\n## Cards that force recall\n\nA flashcard works when the front makes the student produce the back from memory. A card with a question on the front and four options on the back is a recognition task: the student sees the options and picks the one that looks familiar. It feels like knowing and it is not. Students report those cards as easy and then fail the test, which is the check question, and it is the most common failure in generated revision material.\n\nSpecify the format:\n\n> Make flashcards for these topics. Front: a question or a prompt that has one specific answer. Back: the answer only, in at most 15 words. No multiple choice. No yes/no questions. For every definition card, also make a reverse card that gives the definition and asks for the term.\n\nThe reverse card doubles the retrieval and catches the student who has learned the word without the meaning.\n\nAsk for a spread of card types: a term to define, a definition to name, a diagram element to label from a description, a step in a process to supply given the neighbours, a worked calculation with the numbers changed. A set that is all definitions teaches vocabulary and nothing else.\n\n## Free tools that do the spacing\n\nAnki is free on desktop and Android (AnkiDroid), and it schedules each card by how well it was recalled last time: a card the student knew is shown again in days, one they missed in minutes. The iPhone app is paid; the web version is free. Ask the model for the cards as CSV with two columns, front and back, and Anki imports the file directly.\n\nFor a class without devices, the same principle works on paper: three envelopes, labelled daily, every three days and weekly. A card the student gets right moves to the next envelope; a card they miss goes back to daily. It is the Leitner system, it is a century old, and it works.\n\n## A spaced schedule for a term\n\nThe model can also draft the spacing across a term. Paste the syllabus for the term and the test date:\n\n> Build a revision calendar. Each topic is first practised the day after it is taught, then after 3 days, then after 10 days, then after 3 weeks. Show which topics fall on each week. Keep any single day under 20 minutes.\n\nCheck it against the school calendar (festivals, sports days, the week the exams are in another room) because the model does not know your term. Then give students the calendar. The schedule is the intervention; the cards are the material.\n\n## Retrieval in the lesson itself\n\nThe cheapest use of all this is the first five minutes of each lesson. Three questions on the board from last week and last month, answered on paper, not discussed until everyone has written. The model generates a term's worth of these in one go from your topic list (\"ninety questions, three per lesson, each drawing on a topic from one to four weeks earlier\") and you have a starter for every lesson that does more for memory than any revision week.\n\n## Past papers and mark schemes\n\nStudents want past-paper practice. The model can produce questions in the style of your board if shown a real paper, and it can explain a real mark scheme in student language, which is useful: \"Rewrite this marking guide as advice to a student about what a full-mark answer must contain.\"\n\nWhat it cannot do is know the actual past papers, and it will invent them if asked. Use the board's real papers, which are usually free on its website, and use the model to explain the scheme and to generate more questions in the shape.\n\n## The honest limit\n\nRetrieval practice works because it is effortful, and effort is unpleasant. Students given a choice prefer rereading, and highlighting, and recognition cards, because those feel productive and are not. So the material is not enough. Explain the mechanism to the class, that the struggle to recall is the thing building the memory, and most of them, once they have seen it work in one test, will keep doing it. Some will not, and no card format changes that."},"check":{"question":"A teacher generates 200 flashcards for Class 10 chemistry. Students report the cards are \"easy\" and then perform poorly in the test. On inspection most cards have four options on the back and the student picks one. What is the mechanism of the failure?","options":[{"text":"Two hundred cards is far too many for one term; a smaller set of fifty would have been learned more thoroughly by everyone.","correct":false,"tempting":"Volume is not the issue; a well-formed set of 200 is normal for a term, and the students found them easy, not overwhelming."},{"text":"The model generated cards on the wrong topics, so the test covered material that the cards had never actually practised.","correct":false,"tempting":"Possible in general but does not fit the report; the students found the cards easy, which is the signature of recognition, not of topic mismatch."},{"text":"Students did not use the cards often enough, and a proper spaced schedule would have fixed the test results on its own.","correct":false,"tempting":"Scheduling helps only if each retrieval is effortful; more repetitions of a recognition task do not build the memory the test required."},{"text":"The cards tested recognition, which feels like knowing and builds little; a card must make the student produce the answer.","correct":true}]},"prev":{"slug":"slides-and-the-projector-you-sometimes-have","title":"Slides, and the projector you sometimes have","url":"https://addaly.com/learn/ai-for-teachers/slides-and-the-projector-you-sometimes-have"},"next":{"slug":"a-unit-plan-not-just-a-lesson","title":"A unit plan, not just a lesson","url":"https://addaly.com/learn/ai-for-teachers/a-unit-plan-not-just-a-lesson"},"authorship":"Founder and series editor: Dr Mustafa Mun. Written by AI models to a written standard. Exam keys checked blind by a second model: 3,665 of 3,666 agreed. Academic review invited.","review":{"status":"Draft, open for academic review"},"cite":"Addaly (2026) 'Revision: flashcards and spaced practice', in AI for Teachers. Edited by M. Mun. Addaly. Available at: https://addaly.com/learn/ai-for-teachers/revision-flashcards-and-spaced-practice (Accessed: 13 October 2026).","licence":{"name":"CC BY-NC-SA 4.0","url":"https://creativecommons.org/licenses/by-nc-sa/4.0/"},"reader":{"schema":1,"title":"Revision: flashcards and spaced practice","courseTitle":"AI for Teachers","module":{"index":3,"slug":"planning-and-materials","title":"Planning and making materials"},"answer":null,"keyIdea":"The evidence for retrieval practice and spacing is strong and the model can generate a term’s flashcards and a spaced schedule in minutes; the failure is cards that test recognition rather than recall, and the fix is a card format that forces the student to produce the answer.","hero":{"kind":"figure","figure":{"t":"figure","label":"Figure","figure":{"type":"line","title":"Why the effortful version wins later","series":[{"label":"Read the notes three times","points":[[1,79],[7,52],[14,40],[28,29]]},{"label":"Answered from memory, spaced","points":[[1,74],[7,67],[14,62],[28,56]]}],"xLabel":"Days since the lesson","yLabel":"% of items recalled in a test","caption":"Both groups spent the same total time. Rereading feels more productive on the day and is measured lower a month out; producing the answer from memory, spread over days, holds. The shape is the well-replicated one; the exact numbers depend on the material. This is why a card with four options on the back teaches recognition and not recall."},"alt":"Figure: Why the effortful version wins later. Line chart of % of items recalled in a test against Days since the lesson: Read the notes three times (1, 79) (7, 52) (14, 40) (28, 29); Answered from memory, spaced (1, 74) (7, 67) (14, 62) (28, 56). Both groups spent the same total time. Rereading feels more productive on the day and is measured lower a month out; producing the answer from memory, spread over days, holds. The shape is the well-replicated one; the exact numbers depend on the material. This is why a card with four options on the back teaches recognition and not recall."}},"blocks":[{"t":"h2","id":"two-findings-that-survive-every-replication","runs":[{"text":"Two findings that survive every replication"}]},{"t":"p","runs":[{"text":"Two results from the research on memory hold up wherever they are tested. Retrieval practice, trying to produce an answer from memory, builds lasting memory far better than rereading, even when the rereading feels more productive. Spacing, the same material returned to over days and weeks, beats the same total time spent in one block. Neither needs technology. Both are made much easier by it."}]},{"t":"p","runs":[{"text":"The model’s role is to generate the material for retrieval at volume. Your role is to make sure the material actually forces retrieval, which is the thing it gets wrong."}]},{"t":"h2","id":"cards-that-force-recall","runs":[{"text":"Cards that force recall"}]},{"t":"p","runs":[{"text":"A flashcard works when the front makes the student produce the back from memory. A card with a question on the front and four options on the back is a recognition task: the student sees the options and picks the one that looks familiar. It feels like knowing and it is not. Students report those cards as easy and then fail the test, which is the check question, and it is the most common failure in generated revision material."}]},{"t":"p","runs":[{"text":"Specify the format:"}]},{"t":"quote","runs":[{"text":"Make flashcards for these topics. Front: a question or a prompt that has one specific answer. Back: the answer only, in at most 15 words. No multiple choice. No yes/no questions. For every definition card, also make a reverse card that gives the definition and asks for the term."}]},{"t":"p","runs":[{"text":"The reverse card doubles the retrieval and catches the student who has learned the word without the meaning."}]},{"t":"p","runs":[{"text":"Ask for a spread of card types: a term to define, a definition to name, a diagram element to label from a description, a step in a process to supply given the neighbours, a worked calculation with the numbers changed. A set that is all definitions teaches vocabulary and nothing else."}]},{"t":"h2","id":"free-tools-that-do-the-spacing","runs":[{"text":"Free tools that do the spacing"}]},{"t":"p","runs":[{"text":"Anki is free on desktop and Android (AnkiDroid), and it schedules each card by how well it was recalled last time: a card the student knew is shown again in days, one they missed in minutes. The iPhone app is paid; the web version is free. Ask the model for the cards as CSV with two columns, front and back, and Anki imports the file directly."}]},{"t":"p","runs":[{"text":"For a class without devices, the same principle works on paper: three envelopes, labelled daily, every three days and weekly. A card the student gets right moves to the next envelope; a card they miss goes back to daily. It is the Leitner system, it is a century old, and it works."}]},{"t":"h2","id":"a-spaced-schedule-for-a-term","runs":[{"text":"A spaced schedule for a term"}]},{"t":"p","runs":[{"text":"The model can also draft the spacing across a term. Paste the syllabus for the term and the test date:"}]},{"t":"quote","runs":[{"text":"Build a revision calendar. Each topic is first practised the day after it is taught, then after 3 days, then after 10 days, then after 3 weeks. Show which topics fall on each week. Keep any single day under 20 minutes."}]},{"t":"p","runs":[{"text":"Check it against the school calendar (festivals, sports days, the week the exams are in another room) because the model does not know your term. Then give students the calendar. The schedule is the intervention; the cards are the material."}]},{"t":"h2","id":"retrieval-in-the-lesson-itself","runs":[{"text":"Retrieval in the lesson itself"}]},{"t":"p","runs":[{"text":"The cheapest use of all this is the first five minutes of each lesson. Three questions on the board from last week and last month, answered on paper, not discussed until everyone has written. The model generates a term’s worth of these in one go from your topic list (“ninety questions, three per lesson, each drawing on a topic from one to four weeks earlier”) and you have a starter for every lesson that does more for memory than any revision week."}]},{"t":"h2","id":"past-papers-and-mark-schemes","runs":[{"text":"Past papers and mark schemes"}]},{"t":"p","runs":[{"text":"Students want past-paper practice. The model can produce questions in the style of your board if shown a real paper, and it can explain a real mark scheme in student language, which is useful: “Rewrite this marking guide as advice to a student about what a full-mark answer must contain.”"}]},{"t":"p","runs":[{"text":"What it cannot do is know the actual past papers, and it will invent them if asked. Use the board’s real papers, which are usually free on its website, and use the model to explain the scheme and to generate more questions in the shape."}]},{"t":"h2","id":"the-honest-limit","runs":[{"text":"The honest limit"}],"role":"limit"},{"t":"p","runs":[{"text":"Retrieval practice works because it is effortful, and effort is unpleasant. Students given a choice prefer rereading, and highlighting, and recognition cards, because those feel productive and are not. So the material is not enough. Explain the mechanism to the class, that the struggle to recall is the thing building the memory, and most of them, once they have seen it work in one test, will keep doing it. Some will not, and no card format changes that."}]}],"check":{"question":[{"text":"A teacher generates 200 flashcards for Class 10 chemistry. Students report the cards are “easy” and then perform poorly in the test. On inspection most cards have four options on the back and the student picks one. What is the mechanism of the failure?"}],"options":[{"text":[{"text":"Two hundred cards is far too many for one term; a smaller set of fifty would have been learned more thoroughly by everyone."}],"correct":false,"tempting":[{"text":"Volume is not the issue; a well-formed set of 200 is normal for a term, and the students found them easy, not overwhelming."}]},{"text":[{"text":"The model generated cards on the wrong topics, so the test covered material that the cards had never actually practised."}],"correct":false,"tempting":[{"text":"Possible in general but does not fit the report; the students found the cards easy, which is the signature of recognition, not of topic mismatch."}]},{"text":[{"text":"Students did not use the cards often enough, and a proper spaced schedule would have fixed the test results on its own."}],"correct":false,"tempting":[{"text":"Scheduling helps only if each retrieval is effortful; more repetitions of a recognition task do not build the memory the test required."}]},{"text":[{"text":"The cards tested recognition, which feels like knowing and builds little; a card must make the student produce the answer."}],"correct":true,"tempting":null}],"order":[2,1,0,3],"why":null},"prev":{"slug":"slides-and-the-projector-you-sometimes-have","title":"Slides, and the projector you sometimes have","minutes":8},"next":{"slug":"a-unit-plan-not-just-a-lesson","title":"A unit plan, not just a lesson","minutes":8}}}