AI is now part of many trainee pharmacists' revision workflows. Used well, it accelerates understanding and closes knowledge gaps. Used badly, it becomes a crutch that produces the illusion of readiness without the underlying competence the exam actually tests.
The GPhC CRA framework is explicit: the assessment requires trainee pharmacists to apply underpinning pharmacy knowledge and evidence to safe, effective, and professional care within UK practice. It prioritises application over factual recall. The exam tests whether you can make the right clinical decision in a realistic scenario — not whether you can recite a fact from a textbook.
Why AI Is Useful for Pharmacy Revision
AI can explain concepts in multiple ways — if one explanation does not land, the trainee can ask for an alternative using a different analogy or step-by-step breakdown. It can generate comparison tables between similar drugs — "compare ACE inhibitors vs ARBs by mechanism, indications, contraindications, key interactions, monitoring, and counselling points" — helping trainees build the discrimination skills the CRA tests. It can work through calculation steps — showing dimensional analysis, unit conversion pathways, and identifying exactly where a rounding error was introduced. It can answer "why not the other options?" — the reasoning that separates passing from failing in SBA questions, where understanding why four options are wrong is as important as knowing why one is right. It can link topics across domains — connecting pharmacology (mechanism) to therapeutics (clinical use) to law (prescribing restrictions) to patient counselling (what to tell the patient) in ways that fragmented textbook chapters do not always make explicit.
Why AI-Only Revision Is Risky
AI generates explanations. Explanations create comprehension. Comprehension feels like readiness. But the CRA tests retrieval and application under time pressure — not comprehension during passive reading.
Part 1 gives you 40 calculation questions in 120 minutes — 3 minutes per question, free-entry numerical answers, no multiple choice. Reading an AI-generated explanation of displacement volume calculations is fundamentally different from calculating a displacement volume correctly in 3 minutes with a blank answer box.
Part 2 gives you 120 MCQs in 150 minutes — 75 seconds per question average. Reading an AI explanation of why trimethoprim is contraindicated in pregnancy is different from selecting trimethoprim as the contraindicated option from five plausible antibiotics in a clinical scenario where the patient's pregnancy is mentioned once in a dense clinical vignette.
The testing effect — one of the most robust findings in learning science — shows approximately 50% better long-term retention from retrieval practice compared to passive study. AI explanations are passive input. Exam questions are active output. Both have value. Exam performance depends overwhelmingly on the active output.
What the GPhC CRA Actually Tests
Part 1: 40 numerical free-entry calculations in 120 minutes. Concentrations, dilutions, displacement volumes, dose regimens (weight-based, age-adjusted, BSA-based), infusion rates, unit conversions, pharmacokinetics, pharmacoeconomics, quantities to supply. No multiple choice. No option list. The November 2024 pass mark was 28/40 (70%), and nearly 30% of candidates failed Part 1 alone.
Part 2: 120 MCQs in 150 minutes — 90 SBAs and 30 EMQs. Clinical therapeutics (prescribing, monitoring, counselling, adverse effects, interactions, dose adjustments for special populations), law governance and regulation (Medicines Act, Misuse of Drugs Act, Human Medicines Regulations, GPhC Standards, responsible pharmacist regulations, controlled drug requirements), and person-centred care (communication, shared decision-making, safeguarding, confidentiality, consent).
Both parts must be passed in the same sitting. No compensation between parts. Maximum 3 lifetime attempts.
How to Use AI Safely in Revision
Ask for explanations after Q-bank questions. When you get a question wrong, use AI to understand why. This converts a missed question into a learning event.
Generate comparison tables. Drug class comparisons that build discrimination — the skill most directly tested by SBAs.
Test counselling points. Generate counselling scenarios with AI, then verify against the SmPC. Practise both retrieval and verification.
Practise "why not?" After each SBA, ask AI to explain why each distractor is wrong. This builds the discriminatory reasoning the CRA demands.
Turn errors into adaptive targets. Use AI-identified weak areas to focus Q-bank sessions. The iatroX premium pharmacist Q-bank does this automatically — its adaptive engine serves more questions in areas where you make errors.
Do not use AI to generate answers you memorise without understanding. Do not confuse reading AI output with being exam-ready. Do not produce "revision notes" you never test yourself on.
The Revision Stack
Use Ask iatroX to understand the concept — with medicines information powered via eMC/SmPC. Use the iatroX premium pharmacist Q-bank to test whether you can apply it under exam conditions — SBAs, EMQs, and adaptive calculations mapped to the 2026 CRA framework.
Prepare with Ask iatroX for clarification and the premium Q-bank for exam-style practice →