{"course":{"slug":"ai-for-students","title":"AI for Students","url":"https://addaly.com/learn/ai-for-students","level":"beginner","track":"work"},"lesson":{"slug":"exam-hall","title":"Revision That Survives The Exam Hall","url":"https://addaly.com/learn/ai-for-students/exam-hall","position":65,"of":73,"minutes":8,"answer":"Study under the conditions of the test: silent, timed, blank paper, nothing to ask.","keyIdea":"Study under the conditions of the test: silent, timed, blank paper, nothing to ask."},"body":{"format":"markdown","text":"## The mismatch\n\nThe exam hall has no chatbot. It has a desk, a clock, and paper.\n\nIf every hour of your revision happened with a tool open in the next tab, you have practised a task that does not resemble the one being assessed. Not because you cheated, because the conditions were different, and conditions are most of what determines whether something is retrievable.\n\n> Figure: Two different rooms. Where the revision happened: A tab open with something that answers; Typed, with a delete key and a second attempt; Untimed, and paused whenever it got hard; Warm, because you had just read about the topic; Music, a chair you like, a phone within reach. Where the assessment happens: A desk, a clock and paper; Handwritten, and hands cramp at forty minutes; Timed, and the reading time is inside the clock; Cold: you sit down and begin; Coughing, chairs, and nothing at all to consult. This is not a lecture about cheating. Memory is sensitive to the conditions it was formed in, so revision that never resembles the assessment has practised a different task. Closing each of these gaps costs nothing: a kitchen timer, a stack of blank paper, a silent study room and a phone left in another room.\n\nThis lesson is about closing that gap deliberately.\n\n## The core session\n\nOne activity outperforms everything else in the final month, and it is boring.\n\n**Sit a past paper, closed-book, timed, phone in another room.** No lookups until the clock stops. Write by hand if the exam is handwritten: the motor task matters more than people expect, and so does the fact that you cannot delete a sentence and restart it.\n\nThen, and only then, the AI has a job:\n\n```text\nHere is the mark scheme and here is my answer. Mark it\nstrictly. Tell me which marks I lost and why. Do not\nrewrite my answer or show me a better one.\n```\n\nThat final clause is load-bearing. The instant it produces a model answer, you stop diagnosing and start reading, and reading a good answer feels like progress while producing very little of it.\n\nA second prompt worth having:\n\n```text\nBased on what I got wrong, what do I not understand?\nGive me the underlying gap, not the individual errors.\n```\n\nThree lost marks scattered across a paper are usually one misunderstanding wearing three costumes. Finding the costume is easy. Finding the misunderstanding is the job.\n\n## Space it and mix it\n\nTwo findings, both old, both reliable, both ignored every year.\n\n**Spacing.** Five ninety-minute sessions across three weeks beats one fifteen-hour weekend, and it is not close. A workable schedule for any topic: day 1, day 3, day 7, day 14, day 28. Each session is short, cold, and starts with recall before it starts with notes.\n\n**Interleaving.** Do not block. Three hours on integration by parts feels efficient and produces a student who can integrate by parts *when told to*. The exam does not tell you. Mixing problem types (so that each question also requires deciding what kind of question it is) feels worse, scores worse in the session, and transfers substantially better.\n\nAI helps here in one specific way: ask it to build you a mixed problem set drawn across four topics in random order, with no labels. Then close the chat and work through it on paper.\n\n## The lookup log\n\nStarting today, keep one list: **everything you reached for a tool to get.**\n\nEvery formula you couldn't recall. Every date. Every step you needed a hint for. One line each, no self-criticism.\n\nAt revision time, that list is your syllabus. It is a far better guide than any feeling you have about your weak topics, because feelings about weak topics are generated by the same fluency illusion that got you here. The log was recorded at the exact moment of not knowing, which makes it honest.\n\n## What to stop doing\n\nRereading notes. Highlighting. And the new one:\n\n**Having AI condense your notes into a clean summary and studying the summary.** This is now the single most common revision activity among students who have the tools, and it is close to the worst thing you can do with the time. You get someone else's compression, of material you needed to compress yourself, presented with maximum clarity, which produces the maximum feeling of understanding for the minimum retrieval.\n\nIf you want the summary, write it yourself from memory, *then* ask the model what you left out. That sequence is one of the best exercises in this course. The other is a summary you cannot make.\n\n## The test to apply\n\n> If you cannot write it on blank paper, in a silent room, with nothing to consult, you cannot write it in the hall.\n\nRun that check three weeks out, not three days out. Three weeks is enough time to do something about the answer.\n\n## One exception\n\nSome courses now assess with AI permitted: open-tool exams, take-home analyses, coursework where the tools are expected. If that is your assessment, the skill being tested is different: it is judgment, verification, and knowing what to ask. Prepare for *that* by practising with the tools under time pressure, and by practising catching their errors.\n\nBut find out which kind you are sitting. Do not assume."},"check":{"question":"Three weeks before the exam, which revision block is doing the most for performance in the hall?","options":[{"text":"Working past-paper questions with the AI alongside, asking whenever you stall, so you get through many more questions.","correct":false,"tempting":"Coverage feels like progress, but this recreates exactly the unguarded-tutor condition that raised practice scores and lowered exam scores."},{"text":"Sitting a past paper closed-book and timed, then having AI mark it against the scheme without rewriting your answers.","correct":true},{"text":"Having AI condense forty pages of notes into six clear pages, then studying those until everything makes sense.","correct":false,"tempting":"Feels efficient and produces a strong sense of understanding, but the compression was the thinking, and it was done by something else."},{"text":"Having AI generate a hundred new practice questions and reading through them with model answers to see the range of what might come up.","correct":false,"tempting":"Correctly identifies breadth as valuable, but reading answers is not retrieval: the questions only pay off if you attempt them cold."}]},"prev":{"slug":"what-happens-to-what-you-type","title":"What happens to what you type","url":"https://addaly.com/learn/ai-for-students/what-happens-to-what-you-type"},"next":{"slug":"still-worth-learning","title":"The Honest Case For Still Learning It","url":"https://addaly.com/learn/ai-for-students/still-worth-learning"},"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 That Survives The Exam Hall', in AI for Students. Edited by M. Mun. Addaly. Available at: https://addaly.com/learn/ai-for-students/exam-hall (Accessed: 31 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 That Survives The Exam Hall","courseTitle":"AI for Students","module":{"index":8,"slug":"assessment-and-after","title":"Assessment, and the years after it"},"answer":null,"keyIdea":"Study under the conditions of the test: silent, timed, blank paper, nothing to ask.","hero":{"kind":"figure","figure":{"t":"figure","label":"Figure 8.1","figure":{"left":{"title":"Where the revision happened","items":["A tab open with something that answers","Typed, with a delete key and a second attempt","Untimed, and paused whenever it got hard","Warm, because you had just read about the topic","Music, a chair you like, a phone within reach"]},"type":"compare","right":{"title":"Where the assessment happens","items":["A desk, a clock and paper","Handwritten, and hands cramp at forty minutes","Timed, and the reading time is inside the clock","Cold: you sit down and begin","Coughing, chairs, and nothing at all to consult"]},"title":"Two different rooms","caption":"This is not a lecture about cheating. Memory is sensitive to the conditions it was formed in, so revision that never resembles the assessment has practised a different task. Closing each of these gaps costs nothing: a kitchen timer, a stack of blank paper, a silent study room and a phone left in another room."},"alt":"Figure: Two different rooms. Where the revision happened: A tab open with something that answers; Typed, with a delete key and a second attempt; Untimed, and paused whenever it got hard; Warm, because you had just read about the topic; Music, a chair you like, a phone within reach. Where the assessment happens: A desk, a clock and paper; Handwritten, and hands cramp at forty minutes; Timed, and the reading time is inside the clock; Cold: you sit down and begin; Coughing, chairs, and nothing at all to consult. This is not a lecture about cheating. Memory is sensitive to the conditions it was formed in, so revision that never resembles the assessment has practised a different task. Closing each of these gaps costs nothing: a kitchen timer, a stack of blank paper, a silent study room and a phone left in another room."}},"blocks":[{"t":"h2","id":"the-mismatch","runs":[{"text":"The mismatch"}]},{"t":"p","runs":[{"text":"The exam hall has no chatbot. It has a desk, a clock, and paper."}]},{"t":"p","runs":[{"text":"If every hour of your revision happened with a tool open in the next tab, you have practised a task that does not resemble the one being assessed. Not because you cheated, because the conditions were different, and conditions are most of what determines whether something is retrievable."}]},{"t":"p","runs":[{"text":"This lesson is about closing that gap deliberately."}]},{"t":"h2","id":"the-core-session","runs":[{"text":"The core session"}]},{"t":"p","runs":[{"text":"One activity outperforms everything else in the final month, and it is boring."}]},{"t":"p","runs":[{"text":"Sit a past paper, closed-book, timed, phone in another room.","b":true},{"text":" No lookups until the clock stops. Write by hand if the exam is handwritten: the motor task matters more than people expect, and so does the fact that you cannot delete a sentence and restart it."}]},{"t":"p","runs":[{"text":"Then, and only then, the AI has a job:"}]},{"t":"code","lang":"text","label":"Text","text":"Here is the mark scheme and here is my answer. Mark it\nstrictly. Tell me which marks I lost and why. Do not\nrewrite my answer or show me a better one."},{"t":"p","runs":[{"text":"That final clause is load-bearing. The instant it produces a model answer, you stop diagnosing and start reading, and reading a good answer feels like progress while producing very little of it."}]},{"t":"p","runs":[{"text":"A second prompt worth having:"}]},{"t":"code","lang":"text","label":"Text","text":"Based on what I got wrong, what do I not understand?\nGive me the underlying gap, not the individual errors."},{"t":"p","runs":[{"text":"Three lost marks scattered across a paper are usually one misunderstanding wearing three costumes. Finding the costume is easy. Finding the misunderstanding is the job."}]},{"t":"h2","id":"space-it-and-mix-it","runs":[{"text":"Space it and mix it"}]},{"t":"p","runs":[{"text":"Two findings, both old, both reliable, both ignored every year."}]},{"t":"p","runs":[{"text":"Spacing.","b":true},{"text":" Five ninety-minute sessions across three weeks beats one fifteen-hour weekend, and it is not close. A workable schedule for any topic: day 1, day 3, day 7, day 14, day 28. Each session is short, cold, and starts with recall before it starts with notes."}]},{"t":"p","runs":[{"text":"Interleaving.","b":true},{"text":" Do not block. Three hours on integration by parts feels efficient and produces a student who can integrate by parts "},{"text":"when told to","i":true},{"text":". The exam does not tell you. Mixing problem types (so that each question also requires deciding what kind of question it is) feels worse, scores worse in the session, and transfers substantially better."}]},{"t":"p","runs":[{"text":"AI helps here in one specific way: ask it to build you a mixed problem set drawn across four topics in random order, with no labels. Then close the chat and work through it on paper."}]},{"t":"h2","id":"the-lookup-log","runs":[{"text":"The lookup log"}]},{"t":"p","runs":[{"text":"Starting today, keep one list: "},{"text":"everything you reached for a tool to get.","b":true}]},{"t":"p","runs":[{"text":"Every formula you couldn’t recall. Every date. Every step you needed a hint for. One line each, no self-criticism."}]},{"t":"p","runs":[{"text":"At revision time, that list is your syllabus. It is a far better guide than any feeling you have about your weak topics, because feelings about weak topics are generated by the same fluency illusion that got you here. The log was recorded at the exact moment of not knowing, which makes it honest."}]},{"t":"h2","id":"what-to-stop-doing","runs":[{"text":"What to stop doing"}]},{"t":"p","runs":[{"text":"Rereading notes. Highlighting. And the new one:"}]},{"t":"p","runs":[{"text":"Having AI condense your notes into a clean summary and studying the summary.","b":true},{"text":" This is now the single most common revision activity among students who have the tools, and it is close to the worst thing you can do with the time. You get someone else’s compression, of material you needed to compress yourself, presented with maximum clarity, which produces the maximum feeling of understanding for the minimum retrieval."}]},{"t":"p","runs":[{"text":"If you want the summary, write it yourself from memory, "},{"text":"then","i":true},{"text":" ask the model what you left out. That sequence is one of the best exercises in this course. The other is a summary you cannot make."}]},{"t":"h2","id":"the-test-to-apply","runs":[{"text":"The test to apply"}]},{"t":"quote","runs":[{"text":"If you cannot write it on blank paper, in a silent room, with nothing to consult, you cannot write it in the hall."}]},{"t":"p","runs":[{"text":"Run that check three weeks out, not three days out. Three weeks is enough time to do something about the answer."}]},{"t":"h2","id":"one-exception","runs":[{"text":"One exception"}]},{"t":"p","runs":[{"text":"Some courses now assess with AI permitted: open-tool exams, take-home analyses, coursework where the tools are expected. If that is your assessment, the skill being tested is different: it is judgment, verification, and knowing what to ask. Prepare for "},{"text":"that","i":true},{"text":" by practising with the tools under time pressure, and by practising catching their errors."}]},{"t":"p","runs":[{"text":"But find out which kind you are sitting. Do not assume."}]}],"check":{"question":[{"text":"Three weeks before the exam, which revision block is doing the most for performance in the hall?"}],"options":[{"text":[{"text":"Working past-paper questions with the AI alongside, asking whenever you stall, so you get through many more questions."}],"correct":false,"tempting":[{"text":"Coverage feels like progress, but this recreates exactly the unguarded-tutor condition that raised practice scores and lowered exam scores."}]},{"text":[{"text":"Sitting a past paper closed-book and timed, then having AI mark it against the scheme without rewriting your answers."}],"correct":true,"tempting":null},{"text":[{"text":"Having AI condense forty pages of notes into six clear pages, then studying those until everything makes sense."}],"correct":false,"tempting":[{"text":"Feels efficient and produces a strong sense of understanding, but the compression was the thinking, and it was done by something else."}]},{"text":[{"text":"Having AI generate a hundred new practice questions and reading through them with model answers to see the range of what might come up."}],"correct":false,"tempting":[{"text":"Correctly identifies breadth as valuable, but reading answers is not retrieval: the questions only pay off if you attempt them cold."}]}],"order":[3,0,2,1],"why":null},"prev":{"slug":"what-happens-to-what-you-type","title":"What happens to what you type","minutes":8},"next":{"slug":"still-worth-learning","title":"The Honest Case For Still Learning It","minutes":9}}}