The ISTQB organization, like thousands of others, reacted to the AI boom and recently released an update to its new syllabus, "ISTQB Testing with Generative AI". In this article, I will try to analyze the reasons why the knowledge in this syllabus could be interesting for you and why you might subsequently be interested in taking the exam. I will rely both on my experience in creating training materials for this module and on my experience in reviewing the original content.

Note: While writing, AI did not help me otherwise (with the exception of the final proofreading and fast translation into English). If you miss that and want to have the content of the syllabus summarized, just upload the syllabus text (optimally in MD format) into any LLM and provide the correct prompt.

To begin with, I must demyth one usual mistake – ISTQB CT-GenAI is NOT about how to test AI applications (that is what ISTQB AI Testing is about, see below), but how to use AI in testing. With the exception of basic concepts such as prompt, context window, or hallucination (which you can find in the first chapter), it does not explain any detailed theory to you, but throws you straight into the water and deals purely with how to use AI during standard testing activities, such as requirements analysis, test design, or test reporting. To be able to do this, you must know how to prompt well (that is Chapter 2) and you must know the risks of AI and regulations, or standards, rules, and laws when using AI (Chapter 3). In addition, you will find a gentle introduction to advanced topics in the syllabus, such as RAG, model fine-tuning, and the use of agents (Chapter 4), and a small final "managerial" chapter dealing with general recommendations for introducing AI into the company's test processes (Chapter 5).

If you like analogies, it's a bit like driving school – nobody explains to you how the engine works, what cams are, or how the gearbox is designed, but they explain to you how to drive a car (prompting), how not to cause a traffic accident (risks and regulations), how you can simplify your driving when using intelligent assistants (RAG, agents), and how to keep a logbook (processes in the company).

Why is ISTQB actually releasing this syllabus? That is probably simple – it is reacting to the current situation in the entire industry. Clients want "AI Augmented Testing", developers are riding the vibe-coding wave, and the entire software development lifecycle (SDLC) is today somewhat of an AI Assisted SDLC. Why do clients want it? Because we live in capitalism and every "owner of the means of production wants to maximize their profit" :-) If you can write 40% more test cases with AI in the same unit of time as without it, you are clearly increasing the profit of your breadwinner (by the way, you will read in the syllabus that it cannot be calculated so simply). AI will simply almost certainly allow you to be better, more effective, more efficient, and higher-performing testers, provided that you know how to use it fully in accordance with the laws and recognize when it is lying to you.

In this situation, knowledge of AI is no longer a competitive advantage for you, but a necessity, and in the future, apparently a matter of course without which nobody will hire you in 5 years. It can be said that knowledge of AI is actually a new essential skill for a tester, just as at least a basic knowledge of test automation became years ago.

For those of you who are now experiencing so-called FOMO (fear of missing out), i.e., panic that this will happen to you, I have perhaps good news: it is not too late, and if you are serious about your professional development, feel free to start with ISTQB. ISTQB CT-GenAI can be exactly the starting point you are looking for.

This brings me to who the syllabus is intended for. I will probably start traditionally with who it is not intended for. If you are testers who:

• know all the principles of correct prompting (components of a prompt, system and user prompt, prompting techniques),
• understand what hallucination, reasoning errors, and bias are, can recognize them and minimize their occurrence,
• have an overview of all standards and regulations in the field of AI,
• have created your own RAGs for standardized project activities (in any environment and LLM),
• have written your own scripts in Python to work with LLMs and automate the entire process,
• have installed or even fine-tuned local models, e.g., in AnyLLM,

then this syllabus is probably no longer for you. For you, I would easily recommend reading it and, if you want to have an official "signature" on your knowledge from a globally known company, going straight to the exam. It is possible that there will be some details in the syllabus that you do not use and do not know the theory for (e.g., calculation of metrics), but you can easily learn that.

For everyone else, the syllabus is suitable. Here are a few tips for the exam (if you do not want to take accredited training):

• Chapter 2 (prompting) is fundamental and key. This chapter contains the largest number of objectives, and thus a large part of the exam is logically focused precisely on it. If you do not master it, with a high probability you will not be able to succeed in the exam.
• Each chapter begins with the definition of the objectives to be achieved, with each objective assigned exactly one so-called K-level. These are defined as K1 (remember), K2 (understand), and K3 (apply). The exam (or its questions) verifies knowledge of these objectives according to K-levels. That means K1 is enough to remember, but K3 must be applied in a specific situation, and thus you must know and understand the given area.
• Each chapter also contains practical objectives (hands-on), and each practical objective is linked to one K2 or K3 objective. Although practical exercises are not part of the exam, performing them is key to understanding the respective K-objectives. Therefore, we recommend testing them realistically on available LLM models.
• Test the conditions, question types, and time limit of the exam using a sample exam (see links at the end of the article). From our experience, this is the best indicator of whether you are able to succeed in the real exam.

P.S. If you are interested in how AI applications are tested, take a look at the syllabus of the AI Testing module. I will write more about it next time.

Finally, a few links:

• Syllabus CT-GenAI v1.1
• Sample Exam
• ISTQB® Glossary
• ISTQB Press Release
• Webinar on ISTQB AI Tester and Testing with Gen-AI