AI mock viva practice works for MRCOG Part 3 when you make it feel like the real station. Rehearse short, pressured tasks; answer out loud; and ask the AI to judge you on patient safety, information gathering, communication, and applied clinical knowledge, not just whether your facts were vaguely acceptable. The current MRCOG Part 3 Clinical Assessment is a 14-task circuit, with each task lasting 12 minutes including two minutes of reading time, so your practice should be brisk, structured and slightly uncomfortable.
Used well, AI gives you repetition. Used badly, it becomes a fancy way to read notes and feel productive.
Why this matters
This exam rewards visible clinical thinking. Candidates do better when examiners can hear a safe opening, a focused line of questioning, a clear recommendation, and a sensible close. If your answer takes three minutes to get to the point, you usually sound less safe than you are.
The format makes that especially important. MRCOG Part 3 uses 14 tasks, with simulated patient or colleague encounters and structured discussions with an examiner; circuits may also include linked tasks and writing tasks. A lay examiner appears on four tasks, so clarity, empathy and patient-centred language are part of the assessment, not a soft extra.
RCOG also publishes the domains you should train against: patient safety, communication with patients and relatives, communication with colleagues, information gathering, and applied clinical knowledge. Those domains make an excellent AI feedback rubric. If your mock-viva tool is not pushing you on those areas, it is rehearsing the wrong exam.
AI should not be your only calibration point. RCOG's own revision resources include examiner-designed courses and eLearning interactive videos built to reflect a 14-task circuit, which makes them useful for checking the tone and standard you are aiming for.
AI mock vivas for MRCOG Part 3: set-up that actually helps
Build your AI stations by interaction type, not by dumping a whole textbook into the prompt. In practice, you need three rehearsal modes: a patient or relative counselling station, a colleague-to-colleague discussion, and a structured discussion with an examiner. Those mirror the official task styles far better than generic "quiz me on obstetrics" prompts.
Because the real task is time-limited, your AI station should be time-limited too. Give yourself two minutes to read, then speak without pausing to edit.
Give the AI a proper brief
A good prompt makes the station tighter and the feedback sharper. Include:
- your role and setting: registrar in labour ward, antenatal clinic, early pregnancy unit, gynaecology ward
- the exact task: counsel, consent, prioritise, hand over, explain results, discuss next steps
- the station type: patient, colleague, or structured discussion
- strict timing: reading time first, then live questioning
- the feedback domains: patient safety, communication, information gathering, applied clinical knowledge
- instructions to interrupt you if you ramble, use jargon, or miss an obvious safety point
A prompt like this is usually enough:
Act as an MRCOG Part 3 examiner. Give me a two-minute reading stem, then run a timed station. I am the registrar. The task is to counsel a 34-year-old with heavy bleeding after miscarriage who is frightened about future fertility. Interrupt if I am unclear or unsafe. After the station, give feedback under these headings: patient safety, communication, information gathering, applied clinical knowledge, and what I should say differently in my first 30 seconds.
Use the syllabus to choose stations
Do not let the AI recycle the same three emergencies. The RCOG syllabus page remains the official topic map, and the linked syllabus document currently in use is the MRCOG Syllabus and Knowledge Requirements for Core Curriculum dated April 2024. Use that to rotate through obstetric, gynaecological, postoperative, early pregnancy, reproductive health, oncology, urogynaecology, clinical-skills and teaching or research-flavoured scenarios so your practice stays broad.
A simple rule helps: for every high-acuity station you practise, do one calmer station that is mostly about explanation, consent or shared decision-making. That is where many candidates drift into vagueness.
How to answer so the AI rewards the right things
Most strong answers have a shape that examiners can hear. You are not trying to sound encyclopaedic. You are trying to sound safe, organised and registrar-level.
For patient or relative stations
Use a four-part pattern:
- Open and orient: introduce yourself, acknowledge concern, set a brief agenda
- Check immediate safety: look for red flags before launching into explanations
- Explain and recommend: give a plain-English summary, then your recommended next step
- Close safely: check understanding, invite questions, safety-net, and mention documentation or team involvement where relevant
For example, if the stem is a 32-year-old at 38 weeks worried about reduced fetal movements and keen to go home, a good opening sounds like this:
Hello, I’m the obstetrics registrar. I can see this is worrying. First I want to make sure you and the baby are safe right now, then I’ll explain what I’m concerned about and what I’d recommend next, and I’ll answer any questions as we go.
That opening buys you three things straight away: empathy, structure and safety.
For colleague or structured discussion stations
Use a different pattern. Think: headline concern, immediate priorities, focused data, plan, escalation.
A registrar-level answer often starts with one clean sentence: what is the problem and why am I concerned? Then move to your priorities. After that, add the key supporting details, not every detail.
For example, in a structured discussion about a 31-week patient with severe headache, hypertension and possible pre-eclampsia, a sharper start is:
I’m concerned about severe pre-eclampsia. My immediate priorities are maternal stabilisation, prompt assessment of fetal wellbeing, and early senior obstetric and anaesthetic involvement while I gather the key investigations and review the clinical picture.
That sounds more convincing than listing facts first and only later revealing why they matter.
Make AI feedback worth listening to
Generic feedback is useless. Ask for feedback under the published domains: patient safety, communication with patients and relatives, communication with colleagues, information gathering, and applied clinical knowledge. That lets the AI tell you where your prioritisation came too late, where your empathy sounded scripted, or where you jumped to management before gathering enough information.
These follow-up questions usually produce better feedback than "How was that?":
- Where did I demonstrate patient safety early, and where did I delay it?
- Which question I asked was low-yield or unfocused?
- Which phrase sounded too technical for a patient-facing station?
- What escalation, documentation or follow-up step did I miss?
- Ask me one harder follow-up question and let me answer it again.
Keep a short debrief log after every station. Three lines are enough:
- one safety point I missed
- one sentence I want to say better next time
- one habit that wasted time
That is the bit people skip. It is also the bit that improves performance fastest.
Common mistakes when using AI for this exam
The exam format itself tells you what can go wrong. Because circuits may include linked tasks and writing tasks, you need to rehearse follow-on reasoning and concise note-making, not just polished first answers. And because four tasks include a lay examiner, brusque phrasing and jargon-heavy counselling are risky habits.
- doing AI practice silently instead of speaking aloud
- practising untimed, then freezing when the station feels fast
- asking for the model answer before you have committed to your own structure
- using AI as a guideline authority instead of a rehearsal partner
- rehearsing only emergencies and neglecting consent, explanation and uncertainty
- ignoring the first 30 seconds and the final 30 seconds of your answer
Practice workflow
Keep the workflow simple enough to survive a busy rota. You do not need marathon sessions. You need regular, deliberate reps.
A weekly loop that works
- Three short AI stations in the week: one patient-facing, one colleague discussion, one structured discussion
- One mini-circuit at the weekend: three or four stations back to back with strict timing
- One human mock every 1–2 weeks: a study partner, supervisor or group session to test spontaneity
- One running error log: repeat the same weak station until the opening is calmer and the plan is clearer
Use official RCOG material to calibrate your standard. The College's revision resources include examiner-designed courses, and its eLearning resource uses interactive videos to mirror a 14-task circuit. A practical loop is watch one official example, run a similar AI station, debrief it, then repeat the same station the next day with a better opening and tighter close.
If you are using Viva Examiner, keep the sequence simple: simulate, speak, score, rewrite your first 30 seconds, then rerun the station. Repetition beats novelty.
Do recheck the RCOG website close to your sitting. The College has an ongoing MRCOG Part 3 Global Expansion and Sustainability Project, with a future-facing exam model planned from 2027, so candidates should confirm live format details rather than rely on an old screenshot or forum post.
Summary
- Make AI practice match the real station pace: brief reading time, then a structured spoken answer.
- Mark yourself against the five published domains, not vague "sounded okay" feedback.
- Practise the official task styles: simulated patient or colleague encounters and structured discussions.
- Use AI for repetition, then check your standard against official RCOG resources.
- Repeat weak stations until your priorities, language and closure sound safe and clear.
References
- https://www.rcog.org.uk/careers-and-training/exams/mrcog-our-specialty-training-exam/mrcog-part-3/mrcog-part-3-format/
- https://www.rcog.org.uk/careers-and-training/exams/mrcog-our-specialty-training-exam/mrcog-part-3/mrcog-part-3-faqs/
- https://www.rcog.org.uk/careers-and-training/exams/mrcog-our-specialty-training-exam/mrcog-part-3/syllabus/
- https://www.rcog.org.uk/media/yeqbkfrw/mrcog-syllabus-and-knowledge-requirements-v10.pdf
- https://www.rcog.org.uk/careers-and-training/exams/mrcog-our-specialty-training-exam/mrcog-part-3/revision-resources/
- https://www.rcog.org.uk/careers-and-training/exams/mrcog-our-specialty-training-exam/mrcog-part-3/mrcog-part-3-key-dates-and-fees/
- https://www.rcog.org.uk/careers-and-training/exams/mrcog-our-specialty-training-exam/mrcog-part-3/global-expansion-and-sustainability-project/