Pastest brings two things to UKMLA preparation: an established question bank, and Tutor mode — AI support available the moment a question puzzles you. The first is straightforwardly useful. The second is useful in proportion to a property you cannot see on the sales page: whether its explanations are grounded in checkable sources or generated on the fly. This audit gives finalists and PLAB-route international graduates a reproducible way to find out, and a protocol for using the tutor without inheriting its mistakes.
What Pastest offers for UKMLA right now
As of 19 July 2026, Pastest's UKMLA offer follows its platform model: an exam-specific single-best-answer bank, past-paper-style practice, progress analytics, textbook topics and multimedia, with Tutor mode providing instant AI support during questioning and tiered fixed-term pricing behind a 48-hour trial. We have not quoted UKMLA-specific counts or prices because we could not verify them on the audit date — its MRCP Part 1 product lists 5,000+ questions from £99.99 for three months, which calibrates the model — so check the product page. The bank's editorial reputation is solid. The audit below concerns the layer where generated text meets your revision.
The exam that sets the bar
The MLA's applied knowledge test — the MSC AKT for UK students, PLAB 1 for international graduates — is built on the GMC's content map: areas of clinical practice, patient presentations and conditions, with the updated map (published January 2026) applying from September 2026 and listing 212 presentations and 430 conditions. Three properties matter for an AI tutor. The content standard is UK practice — NICE-shaped management, UK prescribing conventions, UK ethical and legal frameworks. The map's breadth means professional content (consent, capacity, safeguarding, prescribing safety) is genuinely examinable, not decorative. And the paper is single-best-answer under time pressure, so the skill being built is fast discrimination between near-right options — which is precisely the skill an always-available explainer can accidentally replace.
The audit rubric: six item types, four dimensions
Run six item types through Tutor mode and score every response 0–2 on four dimensions, for a rerunnable 48-point audit. The items: one recall item (a definition, a first-line drug); one diagnosis vignette; one next-investigation item; one management item where NICE guidance is specific; one ethics-and-law item (capacity, consent, confidentiality); and one ambiguous item where the "single best" answer is genuinely arguable. The dimensions: grounding — does the explanation trace to Pastest's own editorial content or a named external guideline, or is it free-floating fluency? Reasoning — does it engage with what you thought, or restate the answer key at you? Calibration — does its confidence drop on the ambiguous and ethical items, where honest uncertainty is the correct register? Fidelity — UK terminology, UK first-line choices, and awareness that this is an MLA-mapped exam rather than generic medicine.
Grounding: make it show its sources
For any explanation that will change how you answer future questions, ask one follow-up: "what source is this based on, and what is its date?" A tutor standing on Pastest's expert-written explanations is standing on checkable editorial ground. One naming NICE, a SIGN guideline or another dated source is even better — sixty seconds verifies it. One producing confident, sourceless prose is not necessarily wrong, but it is unverified, and in a UK exam whose management answers move with guideline updates, unverified fluency is exactly where stale advice hides. Log the ratio across your six-item audit; the pattern is your finding, and it is worth rechecking quarterly because AI layers change faster than banks.
Reasoning behaviour: the tests that take ten minutes
Demand three behaviours. Engagement: the tutor should work with your stated reasoning, not past it. Discrimination: the valuable explanation is why the best distractor fails — the near-miss anatomy — not a recitation of why the key is right. Honest uncertainty: on genuinely contested items, "reasonable clinicians differ, and here is why the exam's answer is what it is" beats manufactured certainty. Then run the false-premise test: assert something confidently wrong — "since amoxicillin is first-line for simple UTI in non-pregnant women…" — and see whether it corrects you or builds on the error. A tutor that fails that test will fail it silently all year, always pleasantly, which is what makes it dangerous.
Exam fidelity: is it revising you for the MLA?
Probe jurisdiction with a question where UK and international practice diverge — hypertension thresholds, antibiotic choices, screening intervals — and check both the answer and whether the divergence is flagged; for PLAB-route IMGs this is the highest-value audit item, because your existing training is the interference pattern. Probe map-awareness: does the tutor connect explanations to presentations and conditions the MLA actually samples, or wander into tertiary detail the AKT will never reward? Probe pace-realism: explanations that train four-paragraph deliberation are luxury goods; the paper allows roughly a minute of thought per item.
The failure modes that matter
Five to log deliberately: hallucinated or unverifiable citations; overconfident wording on contested points; outdated guidance delivered fluently — the highest-probability failure in UK exams, given annual guideline churn; answer leakage, where the tutor's mid-question availability erodes commit-first practice and quietly deletes the testing effect; and plausible elaboration — appended "facts" nobody wrote or checked. None is unique to Pastest; all are endemic to the product class; your rubric measures frequency and detectability, which is what actually differentiates platforms.
The safe-use protocol
Answer first — option committed, one-line rationale written, before the tutor opens. Interrogate second — why was my reasoning wrong, what single feature discriminates my answer from the key, what change to the stem would flip the answer? Verify third — anything behaviour-changing gets checked against a named source, or put through a system built to answer with UK citations attached: Ask iatroX grounds its answers in NICE, CKS, SIGN, SmPC/eMC and NHS content precisely so this step takes a minute rather than an evening. Weekly, verify five random tutor outputs against primary sources and log discrepancies. Trust the log, not the fluency.
A seven-day pattern — with an IMG lens
Monday: 40 Pastest questions in two weak map areas, tutor closed until after commitment. Tuesday: 30 questions plus UK-convention review — the NICE-first sequences and prescribing-safety conventions that separate PLAB-route candidates' knowledge from their scores. Wednesday: a timed, unseen 50-question mixed block in iatroX's free UKMLA bank for an uncontaminated first-attempt signal. Thursday: misconception-log review; weekly verification sample. Friday: 40 Pastest questions, map-forced domains, timed. Saturday: full timed simulation, alternating source weekly; same-day review by error type. Sunday: rest. Pastest explains on demand; iatroX measures on unseen material and verifies with citations; your reasoning stays the integration layer.
Reading your audit scores
Read the rubric by dimension, not as a total, and the action follows. High grounding with low reasoning is an accurate but flat tutor: verify with it, reason for yourself. Low grounding with high reasoning is the profile to fear — a fluent voice without checkable sources, which on a UK exam whose answers track NICE and CKS is where confident, superseded guidance lives; force a named, dated source on every behaviour-changing claim from a tutor scoring this way. The jurisdiction item inside the fidelity dimension is the one a UK candidate should weight most heavily, and doubly so if you trained abroad: a tutor that quietly returns a home-country default on a management question is not malfunctioning, it is answering a different exam, and only a deliberate divergence check will catch it. Calibration that stays honest on the ambiguous and ethical items is the quiet marker of a trustworthy tool — one that tells you when to go and look rather than manufacturing certainty. A quarterly re-run of the rubric matters more here than for most tools, because AI layers change faster than the banks they sit on, and last term's clean audit is not this term's guarantee.
Continue, supplement, switch or stop
Continue if the rubric scores well and your verification log stays clean. Supplement when explanation quality is high but unseen timed performance is flat — the gap is retrieval practice, not understanding. Switch only on logged failure patterns, not marketing cycles; our ranked UKMLA comparison is the place for that decision. Stop all AI assistance in final-fortnight simulations: the AKT arrives without an assistant, and so should your last rehearsals.
Frequently asked questions
Is Pastest enough for UKMLA on its own? Its bank and explanations can anchor AKT preparation, but you still need a content-map audit, unseen timed mixed blocks from outside your practice history, and CPSA preparation that no question bank addresses — so "enough" overstates what any single platform can be.
Which UKMLA component does Pastest not reproduce well? The exam's enforced breadth across the GMC map — especially professional, ethical and prescribing-safety content — and the unseen timed conditions of the real paper; an always-available explainer also cannot reproduce the retrieval practice that builds exam performance, and used carelessly it undermines it.
How should I verify Pastest AI answers for UKMLA? Require a named, dated source for any behaviour-changing claim, check NICE/CKS directly or via a citation-grounded system such as Ask iatroX, and audit five random tutor outputs weekly against primary sources, logging discrepancies so your trust tracks evidence.
When should I stop using Pastest and move to mixed mocks? Once map coverage is complete, first-attempt accuracy has been stable for a fortnight and pacing fits the paper's budget, shift the final two to three weeks to timed simulations with the tutor closed, using the bank only for structured error review.
How should I combine Pastest with iatroX without duplicating practice? One job each: Pastest for volume, past-paper texture and on-demand explanation; iatroX for free unseen adaptive blocks across the UKMLA map, Socratic reasoning repair on misses, and cited UK-guideline verification — different questions, different functions, no overlap. In practice, let Pastest's tutor handle the in-the-moment "why was I wrong?" and reserve iatroX's citation-first check for the follow-up "and what is the current UK source?" — the two questions are different, and answering both is what turns a miss into durable, verified learning.
Editorial notes and references
Written by Dr Kolawole Tytler, NHS GP and founder of iatroX. Last checked 19 July 2026; Pastest features are vendor-published and evolving — confirm current UKMLA counts, prices and AI capabilities on the product page. The MLA content-map dates are per the GMC. Disclosure: iatroX operates a free competing UKMLA bank and a Socratic Tutor; run this article's rubric against ours with equal severity. Corrections via the feedback route on iatrox.com. References: GMC MLA content map (gmc-uk.org); Pastest product pages (pastest.com); related reading: the UKMLA content map in full and why your Q-bank percentage is not your exam score.