Learn Better With AI

Methods · Generative practice & retrieval

B3

Learner-generated questions

Write the questions you expect on the exam, then answer them, so you practise the shape of the test, not only the content.

When to use this
While you practise
Learning impact
Strong change
Learning evidence
Credible with caveats
Where you are with the topic
Strengthening memory
How often to use it
Worth repeating often
What AI is doing (1)Create practice, examples, or schedules
What AI is doing (2)Evaluate your work

When to use it

  • You have notes that feel complete, but you have never turned them into questions.
  • An exam is coming and you want to practise the kinds of asks you will face, not just re-read.
  • You keep studying details and skipping the bigger “compare”, “explain why”, and “evaluate” moves.
  • You want the AI to critique your questions for being too shallow or too close to the wording in the notes.

What to ask AI

Type something like: “Here is my topic and level. I will write exam-style questions first. Tell me which are deep enough and fair, then wait while I answer before you mark.”

Ask it to reject trivia that only rewards copying a sentence from the notes.

Keep answering your own questions before you ask for model answers.

How it helps

Writing a good question forces you to decide what would count as understanding.

Answering someone else’s list can stay shallow; inventing the list makes you face the gaps in your map.

The AI’s job is to pressure-test your questions and then wait while you answer, not to hand you a past paper and the mark scheme at once.

How to run it

  1. Name the exam format (short answer, essay, problem set) and the topic boundary.
  2. Write five to eight questions yourself before asking for help.
  3. Ask the AI to flag questions that are too easy, too vague, or too close to note wording.
  4. Revise the list, then answer under timed or closed-book conditions.
  5. Only after your answers, ask for feedback on reasoning, not a rewrite of your whole script.

Example exchange

You

A-level biology, photosynthesis. I drafted three questions: (1) Define NADPH. (2) Explain why the Calvin cycle needs ATP and NADPH. (3) Compare light and dark reactions. Can you tell me which are deep enough for an exam, then wait while I answer before you mark?

AI

Question 1 is mostly vocabulary, raise it or drop it. Question 2 is strong. Question 3 works if you force a clear compare criterion, like inputs, location, and products. Send your answers to 2 and 3 when you are ready.

You

For question 2: ATP supplies energy for the carbon-fixation steps, and NADPH supplies the electrons and hydrogen for reducing PGA toward sugar.

AI

Solid start. One focused question: where exactly is ATP used in the cycle, and what happens if NADPH runs out mid-turn? Answer that before I add anything.

Copyable prompt

Topic: [ ]. Level/format: [ ].
I will write exam-style questions first. Critique each for depth, fairness, and
whether it only rewards copying notes. Do not give model answers until I submit
my attempts.

The Tell

You will know the method has flipped when the AI writes the whole question bank and you only recognise good questions instead of inventing them.

Then you practised taste for exam style, not the harder skill of deciding what understanding looks like.

Delete its list, write five of your own from a blank page, and only then ask for a critique.

Principle evidence

Strength of the underlying learning idea, not a claim about AI products.

Generating questions is a generative learning move: you organise material around what would count as a test of understanding. That sits alongside retrieval practice when you then answer your own items.

AI delivery evidence

Whether an AI tutor delivers this method well is a separate question.

AI is good at polishing question wording and spotting trivia. It is not a validated exam board. Prefer its critique of your list over letting it invent the whole paper unsupervised.

Related methods