When written requirements are given in text documents, the best way to generate test cases is by using Natural Language Processing (NLP). Here's why:
* Natural Language Processing (NLP): NLP can analyze and understand human language. It can be used to process textual requirements to extract relevant information and generate test cases. This method is efficient in handling large volumes of textual data and identifying key elements necessary for testing.
* Why Not Other Options:
* Analyzing source code for generating test cases: This is more suitable for white-box testing where the code is available, but it doesn't apply to text-based requirements.
* Machine learning on logs of execution: This approach is used for dynamic analysis based on system behavior during execution rather than static textual requirements.
* GUI analysis by computer vision: This is used for testing graphical user interfaces and is not applicable to text-based requirements.
References:This aligns with the methodology discussed in the syllabus under the section on using AI for generating test cases from textual requirements.

The correct answer is B. Value change coverage . The CT-AI syllabus defines value-change coverage as requiring test cases to cause each neuron to achieve two activation values , where the difference between the two values exceeds a chosen value.
The question states that the test engineer wants to count neurons that have produced two activation function results with a minimum difference of 0.5 . This is precisely the value-change coverage concept: the selected threshold is 0.5, and a neuron is covered when its activation values vary by at least that threshold. Option A, neuron coverage, only requires neurons to achieve an activation value greater than zero; it does not assess the magnitude of change between two outputs. Option C, sign-change coverage, requires neurons to achieve both positive and negative activation values, which focuses on sign rather than numerical distance. Option D is incorrect because the syllabus explicitly includes a coverage criterion matching the described objective.
References/topics: CT-AI Syllabus Chapter 6, Section 6.2 "Coverage Measures for Neural Networks."
=========

The ISTQB CT-AI syllabus defines these two concepts clearly inSection 2.1 - Flexibility and Adaptability. Flexibility is described as the ability of a system to operate in situationsnot explicitly covered in its original requirements, while adaptability refers to how easily the system can be modified to handle new environments or conditions. The syllabus stresses that both flexibility and adaptability are crucial, particularly inself-learning AI systemsthat may need to respond to changes in their environment and adjust their behavior accordingly. It states that systems must be capable of determining when and how to adjust behavior in evolving situations, especially when the operational environment is not fully known at deployment time . This directly aligns with Option A.

Dynamic testing for inappropriate bias often requires obtaining additional attributes about the data being processed. This is because bias can emerge based on hidden or unexamined features that influence the system's behavior. By collecting and analyzing more attributes, testers can better understand potential biases in the data and how they affect the AI system's decisions.

The correct answer is C , covering examples iii and iv . The CT-AI syllabus defines reward hacking as the situation where an AI-based system achieves a specified goal through a "clever" or "easy" solution that perverts the designer's intent. The goal is technically satisfied, but in a way that defeats the real purpose of the system. The syllabus gives the example of an AI system targeting a highest score by hacking the stored score rather than playing the game properly.
Example iii is reward hacking because destroying the outputs may make the remaining sorting condition trivially true, but it violates the production objective. Example iv is also reward hacking because removing all vulnerable functionality eliminates vulnerabilities by destroying useful system behaviour. Examples i and ii are better categorized as negative side effects : the system pursues a stated goal while ignoring broader quality or stakeholder consequences, such as recovery speed or product quality.
References/topics: CT-AI Syllabus Chapter 2, Section 2.6 "Side Effects and Reward Hacking."
=========