Exam map & study plan

This learning path targets ISTQB Certified Tester AI Testing (CT-AI) v2.0, the current syllabus. It prepares you to test AI-based systems. It is deliberately not a course about using ChatGPT or other generative-AI tools to perform ordinary software testing; ISTQB moved that topic into the separate CT-GenAI certification.

Exam facts to memorize before studying

• Prerequisite: ISTQB Certified Tester Foundation Level (CTFL).
• Questions: 40 multiple-choice questions.
• Total points: 44.
• Pass score: 29 points, equivalent to 65% of the available points.
• Time: 60 minutes.
• Non-native language allowance: 25% additional time where applicable.
• Syllabus version: CT-AI v2.0, General Availability 17 April 2026.
• Self-study: explicitly supported by ISTQB using the syllabus, sample exam, and related material.

Do not convert “29 points” into “29 correct questions.” Some exam items can have different point values; use the official exam structure rather than assuming every question is worth exactly one point.

What is actually examinable

The syllabus learning objectives use cognitive levels K1–K4. Treat them differently:

• K1 — Remember: terminology, definitions, lists, recognition.
• K2 — Understand: explain, distinguish, classify, interpret.
• K3 — Apply: calculate, select, use a technique in a scenario.
• K4 — Analyze: reason across a scenario, compare evidence, identify consequences or the best test approach.

The exam is based on these learning objectives and the syllabus sections supporting them. The hands-on objectives are there to build skill and understanding, but are not directly examined as practical computer tasks. This guide still includes every hands-on area because doing the work makes K2–K4 questions substantially easier.

Official chapter map

The current syllabus has seven chapters. Recommended accredited-training time is 1,170 minutes (19.5 hours) before revision and mock exams:

1. Introduction to Artificial Intelligence — 120 min
2. Quality Characteristics for AI-Based Systems — 45 min
3. Machine Learning — 375 min
4. Testing AI-Based Systems — 195 min
5. Input Data Testing — 180 min
6. Model Testing — 225 min
7. Machine Learning Development Testing — 30 min

The distribution tells you where to spend effort. Machine learning, model testing, system testing, and input-data testing deserve much more revision time than the short final chapter.

Coverage matrix for this guide

Nothing in the seven syllabus chapters is intentionally skipped. The learning path maps them as follows:

• Chapter 1: conventional vs AI systems; narrow/general/super AI; AI technologies; generative AI; AI hardware; developing and hosting models; ML frameworks; regulations and standards → AI foundations.
• Chapter 2: AI-specific quality characteristics; AI and safety; measurable acceptance criteria → AI quality & acceptance.
• Chapter 3: ML forms; workflow; creating a model; pretrained models, fine-tuning and RAG; data preparation; classification metrics; neural networks and neural-network coverage → Machine learning.
• Chapter 4: locked/adaptive systems; statistical testing; oracle problem; GenAI/LLM testing; red teaming; exploratory LLM testing; ML-specific test levels; risk-based strategy → Testing AI systems.
• Chapter 5: input-data risks; bias; pipeline testing; representativeness; constraints; label correctness; hands-on data checks → Input data testing.
• Chapter 6: model risks; model documentation/reviews; probabilistic performance; adversarial testing; metamorphic testing; drift; over/underfitting; A/B; back-to-back → Model testing.
• Chapter 7: ML-development risks and deployment testing → ML development & deployment.

The final sections add hands-on labs, the official ISTQB v2.2 sample exam with 46 published questions, a 40-question original mock exam, and an exam-day review sheet.

Recommended study sequence

Use a three-pass method instead of reading the syllabus repeatedly.

Pass 1 — Build the mental model. Read each chapter in this guide, then skim the matching official syllabus section. Your goal is comprehension, not memorization.

Pass 2 — Make it executable. Complete the practical task at the end of every chapter. For calculations, work without looking at the formula first. For test-design topics, write actual test cases, oracles, properties, and risks.

Pass 3 — Exam mode. Take the official sample exam under time pressure, review every wrong answer against the syllabus, then take the original mock in this guide. A correct guess counts as a weakness until you can explain why the distractors are wrong.

A practical 14-day plan

• Days 1–2: AI foundations + quality characteristics.
• Days 3–6: machine learning, including all metric calculations and neural-network basics.
• Days 7–8: testing AI systems and LLMs.
• Days 9–10: input-data testing.
• Days 11–12: model testing + ML development/deployment.
• Day 13: hands-on labs + terminology review.
• Day 14: official sample exam, original mock exam, targeted revision.

If you already work with ML systems, compress the introductory chapters but do not skip ISTQB terminology. Certification questions often test distinctions that experienced engineers understand informally but name differently.

Readiness rule

You are ready when you can do all four of these without notes:

1. Explain the lifecycle from data acquisition to deployed-model monitoring and name the main test risks at each stage.
2. Calculate and interpret classification metrics from a confusion matrix and choose the metric that matches the business risk.
3. Design tests for a probabilistic or generative system where exact expected outputs are unavailable.
4. Distinguish input-data testing, model testing, component/system testing, deployment testing, and production monitoring.

Practice: create a one-page progress sheet with the seven chapters, K-level weaknesses, lab status, official sample score, and mock-exam score. Re-study by weakness, not by chapter order.