Typical roles
You test a system whose behaviour comes out of training data rather than out of code.
You are accountable for the quality of the data a model is trained on.
You have to justify when a model is good enough to go into production.
ISTQB® CT-AI - AI Testing
A model that scores 94 percent on the test set can still be wrong in production, and wrong systematically. CT-AI covers testing AI-based systems, not using AI to do the testing; CT-GenAI is the module for that. The tecnovy CT-AI training spends 375 of its 1170 minutes of instruction on machine learning alone, so that you know what it is you are examining. The testing then happens on three levels: input data, the model, and the development toolchain.
Attendance certificate4.8/5 on ProvenExpert
What this module delivers.
CT-AI equips you to test systems whose behaviour comes out of data, not code. In the tecnovy training you measure model performance, test input data for bias, and use metamorphic testing where no clear test oracle exists.
How the training runs
- Method
Build models, evaluate them, and make them fail on purpose
- Basis
Official ISTQB syllabus 2.0
- Outcome
40 questions, 29 of 44 points, exam booked separately
Dates & booking
Choose a date that fits
No public dates are scheduled yet.
New dates are published regularly. Join the waitlist or plan an in-house course for your team.
Fit
Who this module is designed for
Prerequisites
Formal prerequisites
For exam admission: the ISTQB Certified Tester Foundation Level certificate.
What helps you in the course
Around six months of hands-on testing, development or data science work, as the syllabus recommends.
Consider instead CT-GenAI · ISTQB® - Certified Tester Testing with Generative AI Training Puts generative AI to work in testing, where CT-AI examines AI-based systems.
Curriculum
ISTQB CT-AI Course in Detail
Syllabus 2.0 splits CT-AI into seven chapters across at least 19.5 hours of instruction. It runs from the fundamentals of AI through machine learning to the three levels the testing happens on: input data, the model, and the development toolchain.
01Introduction to Artificial Intelligence
This chapter settles what exactly the rest of the course is talking about.
- AI-based against conventional systems; narrow, general and super AI
- The different types of AI technology, and generative AI
- Hardware, development and hosting of AI models
- Machine learning development frameworks, regulations and standards
02Quality Characteristics for AI-Based Systems
The short part that fixes what an AI system is measured against at all.
- AI-specific quality characteristics under ISO/IEC 25059, and what sets them apart from classic ones
- AI and safety
- Acceptance criteria for AI-based systems
03Machine Learning
The longest chapter, and the most practical: here you build a model yourself.
- The forms of machine learning, and the machine learning workflow
- Pretrained models, fine-tuning and retrieval-augmented generation
- Data preparation, and calculating the functional performance metrics
- How a deep neural network works, and coverage measures for one
04Testing AI-Based Systems
This is where the case for testing AI systems differently gets made.
- Locked against adaptive AI-based systems
- Why the test approach has to be statistical, and what works as a test oracle
- Testing generative AI, and red teaming
- Test levels and risk-based testing for machine learning systems
05Input Data Testing for Machine Learning Systems
The first of the three levels the testing happens on: what goes into the model.
- Input data risks and the mitigations against them
- Testing for bias and for data representativeness
- Data pipeline testing and dataset constraint testing
- Checking the labels for correctness
06Model Testing for Machine Learning Systems
The second level, and the toolkit for defects with no single correct answer.
- Model risks, model documentation and review
- Functional performance testing of probabilistic systems, and adversarial testing
- Metamorphic testing, A/B testing and back-to-back testing
- Drift testing, and testing for overfitting and underfitting
07Machine Learning Development Testing
The third level: the tooling and the deployment, not the model itself.
- Risks from frameworks and their APIs, and the tests against them
- Installability and rollback testing ahead of deployment
- Canary and shadow testing in production
Outcome
What you will be able to do afterwards
- 01
You tell AI-based systems from conventional ones and place the technology in use.
- 02
You formulate acceptance criteria along the AI-specific quality characteristics defined in ISO/IEC 25059.
- 03
You evaluate model performance with a confusion matrix, precision, recall and F1 score.
- 04
You judge coverage measures for neural networks and what they actually tell you.
- 05
You justify the statistical approach to testing and define test oracles for AI systems.
- 06
You test input data for bias, representativeness and correct labelling.
- 07
You surface model defects with metamorphic testing, A/B tests and back-to-back tests.
- 08
You recognise data drift and concept drift, and plan the testing against them.
Certified Tester AI Testing
Once you pass, you receive the official ISTQB certificate at Specialist Level. It is valid for life and never needs renewing.
Certification terms istqb.org ↗
Examination
ISTQB CT-AI examination
40Questions
- Working time
- 60 min
- Pass mark
- 29 / 44
- Format
- Multiple Choice
- Conditions
- Closed Book
The examination is administered by our exam partner iSQI, not by tecnovy. The syllabus's hands-on exercises are compulsory in the course but explicitly not examinable. If you do not sit the exam in your native language you can apply for 25 percent more time; the application has to be in at least one week before the exam date.
Official examination rules istqb.org ↗Proof of attendance
Certificate of participation
≥80%attendance
The tecnovy certificate of participation records your attendance of the CT-AI training, not a passed ISTQB examination.
Open the Certificate Showroom tecnovy → Offer scope
What the course price does not cover
Not included
- Fee for the ISTQB certification examinationThe exam fee is charged by the exam partner, not by tecnovy: CHF 250 excl. VAT, around EUR 270. The euro amount is converted at the daily rate and shown in the booking step.
Trainers
Why tecnovy
What you get on top with us
01
Certification Guarantee
Our pass rate is impressive. If you do not pass at the first attempt, we cover your second attempt (online exam), provided you sit it within 24 hours of the training.
02
Certificate Showroom
Get your certificate of participation and, if you have one, add your exam certificate from E-Learning. Fully automated, beautifully designed. Just for you, only at tecnovy.
03
Flexible Date Change
If you are not able to attend the course, you can rebook your training free of charge up to one week before the start of the training.
04
Attend Twice, Pay Once
You are welcome to visit the training course online again within a year as a refresher or exam preparation.
05
No Slideshow, Hands-On!
Promised: no PowerPoint marathon. We work in groups, tie theory to practice, and you get real project examples from our experienced trainers plus the exchange with like-minded people.
06
Learn from Experts
We always guarantee you the use of didactically and methodically first-class qualified trainers who draw their knowledge from training experience as well as professional practical and project experience.
FAQs
Frequently asked questions
01What is the difference between CT-AI and CT-GenAI?
CT-GenAI uses generative AI in testing; CT-AI tests AI-based systems. ISTQB drew the line itself: with CT-AI 2.0 the chapter on testing with AI was removed entirely and moved into CT-GenAI. Both modules require CTFL, and neither requires the other.
02Which syllabus version does the CT-AI training run on?
On the English syllabus 2.0 of April 2026, and you sit the exam in English accordingly. There is no German edition of 2.0 yet; the German syllabus is still at version 1.0 and remains valid until 21 October 2027. The two editions differ in structure and in exam points.
03I hold CT-AI 1.0. Do I have to sit the new examination?
No. Your CT-AI certificate from version 1.0 stays valid and never needs renewing. There is no bridging exam: to hold version 2.0 you sit it as a separate examination with the exam provider.
04What do I need in order to sit the CT-AI examination?
The syllabus names the ISTQB Certified Tester Foundation Level certificate as the entry criterion. It additionally and explicitly recommends around six months of hands-on testing, development or data science work, but does not require it. The tecnovy CT-AI training assumes no prior data science knowledge.
05How is the CT-AI examination structured?
The CT-AI exam is 40 multiple-choice questions in 60 minutes, and you pass from 29 of 44 points. Older figures give 120 minutes; ISTQB explicitly recorded that as a correction in its exam tables.
06Is the examination part of the tecnovy CT-AI training?
No. The tecnovy CT-AI training prepares you for the examination, which is administered by our exam partner iSQI. You add it during registration; it is not part of the seminar price.
07Do I have to do calculations in the exam?
No. The syllabus states explicitly that the exam does not require deriving or computing metrics using statistical formulae. You still work through them in the course, because otherwise the metrics cannot be judged at all.
08What do I actually do hands-on in the CT-AI training?
The syllabus prescribes hands-on exercises in four of the seven chapters: you build a machine learning model, prepare data, implement a perceptron, explore a large language model, test input data, and apply metamorphic testing.
09Why does the CT-AI training run for three days?
Because syllabus 2.0 sets a minimum of 19.5 hours of instruction for accredited courses across seven chapters. The machine learning chapter alone takes 375 minutes of that. The tecnovy CT-AI training gives it three days, Monday to Wednesday.
What does your training at tecnovy look like?
CT-AICT-AI - AI Testing
No dates yet
We will let you know as soon as the next date is scheduled.
