How the Exam Works and How to Revise

The exam format

The AI4RE micro-credential exam, as published by IREB and offered through its certification bodies, is:

22 multiple-choice questions.
30 minutes, with 6 additional minutes for candidates not sitting in their native language.
80% to pass. With 22 questions, that means you can afford very few wrong answers.
Unproctored and online.
No formal prerequisites — you do not need CPRE Foundation Level first, though basic RE knowledge is recommended.
Valid for three years, unlike the CPRE Foundation Level certificate, which does not expire.
Available in German and English.

Confirm the current format, fee and conditions with the certification body you register with before booking, as these are set by the certification bodies rather than by this course.

Two implications of that format are worth planning around.

80% of 22 is a demanding bar. Roughly four wrong answers is the difference between passing and not. This is not an exam you pass on general AI familiarity; it rewards knowing the syllabus's specific formulations.

30 minutes for 22 questions is about 80 seconds each. That is enough time to read carefully and answer, but not enough to reason your way to an answer you do not know. Recall has to be fast.

The eleven educational objectives

The syllabus assigns each objective a cognitive level. L1 means know, describe, enumerate, name, recognise, remember. L2 means understand, explain, interpret, summarise, justify, classify, compare — and L2 includes L1.

🔗 Match the Pairs

EO 1.1 Know the variety of AI technologies and their application areasDrop here

EO 1.2 Explain the chatbot pipeline, describe RAG, compare with fine-tuningDrop here

EO 2.1 Know the fundamental concept of how an LLM worksDrop here

EO 2.2 Describe LLMs as probabilistic prediction, not reasoningDrop here

EO 3.1 Explain the importance of context for accurate resultsDrop here

EO 3.2 Name the purpose of prompting patterns, recognise how they differ from techniquesDrop here

EO 4.1 Name typical risks of AI use in REDrop here

EO 4.2 Explain the evolving role of the Requirements EngineerDrop here

EO 5.1 Name typical use cases across the four core RE activitiesDrop here

EO 5.2 Explain the importance of human curation and the RE's role in valid resultsDrop here

EO 6.1 Remember and identify core AI terminologyDrop here

Six L1 objectives, five L2. The L1 objectives reward memorised lists — the five AI families, the ten risks, the five system context aspects, the two management tasks. The L2 objectives reward being able to say why.

Where the weight sits

The syllabus's recommended durations are the best available signal of emphasis:

📅 Timeline

EU 1 AI Basics0.5 hours

EU 2 Large Language Models0.5 hours

EU 3 Prompt Engineering0.5 hours

EU 4 Risks and Responsibilities1 hour

EU 5 Use Cases for AI in RE2.5 hours

EU 6 AI Terminology0.5 hours

Educational unit 5 is worth as much as everything else combined. If revision time is short, spend it on modules 6, 7 and 8 of this course — the four core RE activities and what human curation means for each.

Note also the instruction attached to unit 6: in the examinations, terminology may be used in the context of RE-related questions. Terminology is not a separate block of definition questions. It is woven through the rest.

The distinctions most often confused

Every one of these has appeared in a chapter quiz because each is a place where careless reading loses a mark:

Outdated knowledge vs model drift — static from deployment, versus dynamic during use.
Patterns vs techniques — structure, versus situational phrasing.
System context vs user context — fixed and only complementable, versus supplied per conversation.
What embeddings do vs what LLMs do — locate and compare, versus generate.
When transparency labelling applies vs when it does not — public interest content, and the human-editorial-responsibility exemption.
Formulation vs transformation — structuring in standard formats, versus moving across representations and audiences.

A study plan

If you have a full day. Work modules 1 to 9 in order, taking each chapter quiz immediately after its module. Then sit mock 1 cold. Reread the modules behind anything you got wrong, then sit mock 2. Leave mock 3 for the day before the exam.

If you have two hours. Read the key takeaway and flashcards in every module — they are written to be the compressed version. Then sit mock 1 and use your wrong answers to choose what to read in full.

If you have thirty minutes. Module 6, 7 and 8 flashcards, the ten risks from module 4, and the terminology in module 9.

Using the mocks properly

The three mock exams are diagnostics, not practice runs. Get the most from them by:

Sitting them under real conditions. 30 minutes, no notes, no rereading.
Reading every explanation, including for questions you answered correctly — a right answer for the wrong reason will not repeat.
Mapping each wrong answer to its module, and rereading that module rather than the question.
Spacing them out. Three mocks in one sitting measures your stamina. Three mocks across three days measures your retention.

Each mock covers all six educational units in roughly the proportion the syllabus's durations imply, so a low score in one area points at a specific module.

💡Key Takeaway

The exam is 22 multiple-choice questions in 30 minutes at an 80% pass mark, unproctored, with no prerequisites, valid for three years. Six educational objectives are L1 (recall lists) and five are L2 (explain why). Unit 5 — AI across the four core RE activities — carries as much weight as the other five units combined, and terminology is tested inside RE questions rather than as standalone definitions.