Sign in

Sign in to save your questions and track your progress.

Don't have an account? Create one

AKT writing

Write your own AKT questions, get AI feedback, and build a personal question bank you can practise from.

Why write your own AKT questions?

Writing your own multiple-choice questions is one of the more underused study techniques for the MRCGP. The cognitive science is reasonably robust — the generation effect describes how material you produce yourself tends to be recalled more reliably than material you only read. Writing a five-option MCQ with per-option reasoning also forces you to articulate why each plausible-looking wrong answer is wrong, which is the discrimination the AKT actually tests.

Mosaic adds an AI feedback layer. You write a question; the model rates it for clinical accuracy, distractor quality, alignment to the RCGP curriculum, and likeness to AKT examiner style, then tells you specifically what could be improved. The aim isn't to publish polished items — it's to surface what you don't yet know clearly enough to test.

Quick test

Take a 10-question quiz drawn from the community or your own question bank. Test yourself, track your score, and see how well you know the material.

Community question bank

Browse and practise questions created by other registrars. See how peers approach different topics, read AI feedback, and join the discussion.

AI feedback and AI-generated practice questions are for learning purposes only. Nothing here is a validated RCGP exam question. Always verify clinical information against current guidelines. AI can make errors with clinical content — treat all feedback as a starting point for further study, not a definitive source.
Not affiliated with or endorsed by the RCGP or NHS England.
Privacy Policy · Terms & Conditions

My question bank

Question review

Important This review was generated by AI. AI can make mistakes with clinical content, particularly around recent guideline changes. Always verify clinical information against current, up-to-date guidelines. This is a learning exercise, not a validated assessment tool.

Quick test

Take a 10-question quiz drawn from existing questions. Choose your question source and filter below.

Select all Clear
Question 1 of 10 Score: 0/0

Test Complete

Community question bank

Browse questions shared by other registrars. Read AI feedback, see how peers approach different topics, and join the discussion.

Create a question

Important This tool uses AI to review your question. AI can make mistakes with clinical content, particularly around recent guideline changes. Always verify clinical information against current guidelines. This is a learning exercise, not a validated assessment tool.

Writing your own exam questions is one of the most effective ways to prepare for the AKT. It forces you to think about what's being tested and why, to consider plausible alternatives, and to engage with reasoning at a deeper level than simply answering someone else's questions.

1

Choose what to test

Pick a topic and decide on the single clinical decision point, statistical concept, or organisational principle you want to test. The best questions test one thing clearly, not several things at once.

2

Write the stem

Set the scene with enough information to reach the answer. For clinical questions, write a realistic primary care vignette with relevant demographics, history, and findings. For evidence-based practice questions, present a study or dataset. For organisational questions, describe a practice scenario or ethical dilemma. Keep it focused — unnecessary detail is a distraction, not a challenge.

3

Write a clear lead-in

Ask for the single best answer. Good lead-ins include: "What is the most appropriate next step in management?", "What is the most likely diagnosis?", "Which investigation should be performed first?", "What is the number needed to treat?" Avoid vague phrasing like "Which of the following is true?"

4

Write five options with strong distractors

One correct answer, four distractors. The distractors are where the real learning happens — they should be plausible enough that someone without full understanding might choose them. When you struggle to write good distractors, that often reveals a gap in your own knowledge.

Clinical questions 80% of the exam

Vignettes set in UK primary care. Test diagnosis, investigation, management, prescribing, prevention, or screening decisions.

Distractor ideas A management step from old guidelines that has since changed. A treatment for a similar but different condition. An investigation that's useful but not first-line. A drug from the right class but the wrong choice for this patient.

Evidence-based practice 10% of the exam

Test understanding of study design, critical appraisal, sensitivity and specificity, PPV and NPV, NNT and NNH, absolute vs relative risk, confidence intervals, bias, audit, and levels of evidence.

Distractor ideas Confusing sensitivity with specificity or PPV with NPV. Offering NNT when the question asks for absolute risk reduction. Suggesting the wrong study design for a given research question. Mixing up intention-to-treat with per-protocol analysis.

Organisation and management 10% of the exam

Test knowledge of NHS structures, CQC, GMC duties, consent and capacity (including Mental Capacity Act and Gillick competence), safeguarding, confidentiality, complaints, significant event analysis, sickness certification, DVLA reporting, and controlled drugs.

Distractor ideas Confusing who has legal responsibility in a safeguarding scenario. Mixing up Fraser and Gillick. Offering an action that sounds right but breaches confidentiality. Suggesting a step that requires consent when it's a statutory obligation.
5

Check your answer

Ask yourself: could I defend this answer with reference to a current UK guideline, a statistical principle, or a legal framework? If you're not sure, that's fine — one of the most valuable things this tool does is flag when your intended answer might not be right. That's a learning moment, not a failure.

Why this works Research suggests that students who generate their own questions develop stronger understanding than those who only answer them. By constructing a question, you're practising retrieval, elaboration, and self-assessment simultaneously. You're also learning to identify what you don't know — the gaps become visible when you try to write plausible distractors on a topic you haven't fully mastered.
Write your scenario and question. This could be a clinical vignette, a description of a study or dataset, or a practice management situation — depending on the type of question you're writing.
Write 5 options, select which is correct, and explain your reasoning for each — why is the correct answer best, and why is each distractor plausible but wrong? Forcing yourself to defend every option is where the deepest learning happens.
A
B
C
D
E
Which SLP session(s) does this question relate to? Choose from the list — you can't create new modules here. Adding modules makes your question discoverable in the platform view for each SLP session.
Cross-cutting threads, e.g. prescribing-safety, red-flags, health-inequalities. Type to filter existing tags or create a new one. Press Enter or comma to add; Backspace removes the last chip when the box is empty.
Anything else you want the AI reviewer to know — why you chose the topic, what you're trying to test, areas you're unsure about. The per-option reasoning above is the heart of the exercise; this is just for any wider context.
AI feedback and AI-generated practice questions are for learning purposes only. Nothing here is a validated RCGP exam question. Always verify clinical information against current guidelines. AI can make errors with clinical content — treat all feedback as a starting point for further study, not a definitive source.
Not affiliated with or endorsed by the RCGP or NHS England.
Privacy Policy · Terms & Conditions

Generating your questions...

Verify before relying This review is AI-generated. The clinical accuracy check may itself contain errors, particularly regarding recent guideline changes. Always verify clinical claims against current guidelines.

Save as flashcard

Turn this question into a single-fact flashcard you can revise later. Write the front and back yourself — the synthesis is where the learning happens. The reference pane on the right shows the question and AI review for context.

Area, topic, tags and modules carried over from the question.
The cue the learner sees first. Often a question or short concept.
The fact you want to be able to retrieve. Markdown supported.
Choose one of your personal decks. If you don't have one yet, create it in Flashcards first.

Session Summary