ISTQB ISTQB-CT-AI Online Practice Questions and Exam Preparation

ISTQB ISTQB-CT-AI Online Questions & Answers

• Question 61:

  Which statement about testing to prevent data poisoning and adversarial attacks is correct?

  A. Regression testing can be used to verify data sourcing policies to ensure the source of training data.
  B. The adversarial examples identified during adversarial testing must not be added to the training data so that they do not poison the model.
  C. Adversarial testing consists of using adversarial attacks to identify vulnerabilities so that they can be eliminated.
  D. Using AIB testing to identify data poisoning can better identify outliers than exploratory data analysis.
  C. Adversarial testing consists of using adversarial attacks to identify vulnerabilities so that they can be eliminated.

  ---

  The ISTQB CT-AI syllabus explains in Section 4.5 - Testing AI-Specific Risks that adversarial testing is a structured test activity in which testers apply adversarial attacks --- crafted or perturbed inputs --- to intentionally expose weaknesses in the machine learning model. The purpose is to identify vulnerabilities that could be exploited through data poisoning, evasion attacks, or input manipulation. Option C correctly reflects this syllabus definition: adversarial testing is about using attacks to locate weaknesses so that they can be removed or mitigated.

  Option A is incorrect because regression testing does not verify data sourcing policies; it verifies unchanged functionality after modifications. Option B is incorrect because adversarial examples are often added to training data sets to improve robustness (a practice known as adversarial training), rather than being excluded. Option D is incorrect because AIB testing is not described as superior to exploratory data analysis in identifying outliers; both approaches serve different purposes, and exploratory data analysis remains essential for data quality assessment.

  Thus, Option C is consistent with the syllabus-defined purpose of adversarial testing.

• Question 62:

  In the near future, technology will have evolved, and AI will be able to learn multiple tasks by itself without needing to be retrained, allowing it to operate even in new environments. The cognitive abilities of AI are similar to those of a child aged 1-2 years. In the above quote, which ONE of the following options is the correct name for this type of AI?

  A. Technological singularity
  B. Narrow AI
  C. Super AI
  D. General AI
  D. General AI

  ---

  Technological singularity refers to a hypothetical future point at which AI surpasses human intelligence and is able to improve itself autonomously at an accelerating rate. This concept goes far beyond the capabilities described in the quote.

  Narrow AI (also called weak AI) is designed to perform a specific task or a limited set of tasks. It cannot transfer learning across domains or operate in new environments without retraining, which contradicts the description.

  Super AI refers to an AI that exceeds human intelligence across all cognitive domains. This level of intelligence is significantly more advanced than the child-like cognitive abilities described.

  General AI (also known as strong AI) is defined as an AI system capable of understanding, learning, and applying knowledge across a wide range of tasks and domains, and of adapting to new environments without retraining. This directly aligns with the description in the quote, including the comparison to early human cognitive development.

  Therefore, General AI is the correct answer.

• Question 63:

  Which ONE of the following options does NOT describe an AI-technology-related characteristic that differentiates AI test environments from other test environments?

  A. Challenges resulting from low accuracy of the models.
  B. The challenge of mimicking undefined scenarios generated due to self-learning.
  C. The challenge of providing explainability for the decisions made by the system.
  D. Challenges in the creation of scenarios for human handover in autonomous systems.
  D. Challenges in the creation of scenarios for human handover in autonomous systems.

  ---

  The ISTQB CT-AI syllabus explains that AI test environments differ from traditional test environments mainly due to technology-related characteristics of AI systems. These include model accuracy limitations, the emergence of undefined or unexpected scenarios caused by self-learning behavior, and the need to provide explainability for AI decisions. Options A, B, and C all directly reflect such AI-specific, technology-driven challenges.

  Option D, however, concerns the creation of human handover scenarios in autonomous systems. While this is an important testing concern, it is primarily related to operational, human-machine interaction, and deployment aspects, rather than being an intrinsic AI-technology-related characteristic that differentiates AI test environments themselves.

  Therefore, Option D is the correct answer.

• Question 64:

  Which of the following descriptions of quality aspects of a data set is correct?

  A. The quality aspect "Incomplete data" describes the fact that data is missing, for example, for a certain time interval.
  B. The quality aspect "Data not preprocessed" describes the fact that the collected data was recorded incorrectly.
  C. The quality aspect "Irrelevant data" describes the fact that irrelevant data does not affect the ML model.
  D. The quality aspect "Unbalanced data" describes the fact that the data used should be as up-to-date as possible.
  A. The quality aspect "Incomplete data" describes the fact that data is missing, for example, for a certain time interval.

  ---

  The ISTQB CT-AI syllabus describes several data quality aspects that affect machine learning performance. In Section 2.2 - Data Preparation, it explains that data sets may suffer from issues such as incomplete data, irrelevant data, incorrect data, unbalanced data, or data lacking preprocessing. "Incomplete data" refers to situations in which parts of the required data are missing, for example because certain time periods, records, or data sources were not captured. This matches Option A exactly.

  Option B is incorrect because "data not preprocessed" refers to data that has not undergone steps such as cleaning, normalization, or transformation, rather than data that was recorded incorrectly. Option C is incorrect because irrelevant data does negatively affect machine learning models by introducing noise and unnecessary features, which can degrade learning performance. Option D is incorrect because "unbalanced data" refers to disproportionate class distributions, not to how recent or up to date the

  data is.

  Thus, Option A is the only statement that correctly matches the syllabus definition of this data quality aspect.

• Question 65:

  Which ONE of the following activities is MOST relevant when you have more than the required amount of data available for training?

  A. Feature selection
  B. Data sampling
  C. Data labeling
  D. Data augmentation
  B. Data sampling

  ---

  The ISTQB CT-AI syllabus explains that data sampling is used to select a representative subset of a larger dataset. When more data is available than required for training, sampling helps reduce the dataset size while preserving its statistical properties, making training more efficient and manageable.

  The other activities are less relevant in this scenario. Feature selection focuses on reducing the number of input features rather than the volume of data. Data labeling is required for supervised learning but does not address excess data. Data augmentation is typically used when there is insufficient data, not when there is an excess.

  Therefore, Data sampling (Option B) is the most relevant activit

• Question 66:

  The activation value output for a neuron in a neural network is obtained by applying a computation to the neuron. Which ONE of the following options BEST describes the inputs used to compute the activation value?

  A. Individual bias at the neuron level, activation values of neurons in the previous layer, and weights assigned to the connections between the neurons.
  B. Activation values of neurons in the previous layer, and weights assigned to the connections between the neurons.
  C. Individual bias at the neuron level, and weights assigned to the connections between the neurons.
  D. Individual bias at the neuron level, and activation values of neurons in the previous layer.
  A. Individual bias at the neuron level, activation values of neurons in the previous layer, and weights assigned to the connections between the neurons.

  ---

  In a neural network, the activation value of a neuron is computed using three components:

  Activation values of neurons in the previous layer, which provide the input signals Weights assigned to the connections, which scale the input signals An individual bias at the neuron level, which shifts the summed input The computation is typically expressed as:

  

• Question 67:

  Which statement describes factors related to test data that make testing AI-based systems difficult?

  A. Using the same implementation for data acquisition by data scientists and testers prevents defect masking.
  B. Creating and managing large amounts of test data can be difficult, especially when it needs to be representative.
  C. The input data must always be the same over time, especially in real-world systems.
  D. Artificially generated data requires legal approval and must be sanitized and encrypted.
  B. Creating and managing large amounts of test data can be difficult, especially when it needs to be representative.

  ---

  Section 2.2 - Data Preparation and Section 4.1 - Challenges in Testing AI-Based Systems describe difficulties in obtaining and managing large, representative data sets. AI-based systems require realistic, diverse, and representative data that reflects real-world variation. The syllabus emphasizes that assembling such data sets is time-consuming, resource-intensive, and often constrained by availability, privacy concerns, or domain complexity. Option B directly corresponds to these documented challenges.

  Option A is incorrect because using the same implementation for data acquisition can lead to defect masking rather than preventing it; the syllabus explicitly warns against this risk. Option C is incorrect because real-world input data naturally changes over time, and the syllabus notes that data drift is normal in operational environments; expecting stable input data contradicts this reality. Option D is incorrect because, while data privacy is important, the syllabus does not state that artificially generated data

  always requires legal approval or that sanitization and encryption are mandatory in all cases.

  Thus, Option B accurately reflects the syllabus-defined difficulties related to test data in AI-based systems.

• Question 68:

  Which characteristic of AI-based systems makes it difficult to ensure they are safe (e.g., not harming humans)?

  A. Determinism
  B. Complexity
  C. Interpretability
  D. Robustness
  B. Complexity

  ---

  The ISTQB CT-AI syllabus lists several characteristics that make it difficult to ensure safety in AI-based systems. Section 2.8 - Safety and AI explicitly names the characteristics that complicate safety assurance: complexity, non-determinism, probabilistic behavior, self-learning, lack of transparency, and lack of robustness. Among these, complexity is a core challenge because modern AI systems---particularly those using deep learning---exhibit highly non-linear behavior, large numbers of parameters, and intricate interactions that are difficult to predict and analyze.

  Option B (Complexity) directly aligns with the syllabus and is therefore correct. Option A (Determinism) is incorrect because determinism is not a typical characteristic of AI-based systems and does not increase safety risk; in fact, non- determinism is the challenge identified in the syllabus. Option C (Interpretability) affects trust and explainability, but the syllabus treats it as a transparency issue rather than the primary factor that makes safety assurance difficult. Option D (Robustness) is a desired quality attribute; it is the lack of robustness, not robustness itself, that contributes to safety concerns.

  Thus, Complexity best reflects the syllabus-defined difficulty in ensuring the safety of AI-based systems.

• Question 69:

  A startup company has implemented a new facial recognition system for a banking application on mobile devices. The application is intended to learn at run time on the device to determine whether the user should be granted access. It also sends feedback over the Internet to the application developers. After deployment, the application caused continuous restarts of the mobile devices. Which of the following is the most likely cause of the failure?

  A. The feedback requires a physical connection and cannot be sent over the Internet.
  B. Mobile operating systems cannot process machine learning algorithms.
  C. The size of the application is consuming too much of the phone's storage capacity.
  D. The training, processing, and diagnostic generation are too computationally intensive for the mobile device hardware to handle.
  D. The training, processing, and diagnostic generation are too computationally intensive for the mobile device hardware to handle.

  ---

  The ISTQB CT-AI syllabus explains that on-device training and continuous learning can require substantial computational resources. When training, inference, and diagnostic generation are all performed locally on constrained hardware such as mobile devices, the workload may exceed CPU, GPU, memory, or thermal limits. This can lead to severe performance degradation, instability, or repeated system restarts.

  The other options are incorrect. Internet feedback does not require a physical connection, modern mobile operating systems can execute ML workloads, and storage size alone would not typically cause continuous restarts.

  Therefore, the most likely cause is that the computational demands exceed the capabilities of the mobile device hardware, making Option D correct.

• Question 70:

  Which of the following is a technique used in machine learning?

  A. Decision trees
  B. Equivalence partitioning
  C. Boundary value analysis
  D. Decision tables
  A. Decision trees

  ---

  Decision trees are a fundamental technique used in supervised machine learning. The ISTQB CT-AI syllabus describes a decision tree as a tree-like machine learning model in which internal nodes represent decisions, branches represent possible outcomes, and leaf nodes represent predictions or classifications. Decision trees are commonly used for both classification and regression tasks.

  The other options are software testing techniques rather than machine learning techniques. Equivalence partitioning and boundary value analysis are black-box test design techniques, and decision tables are used to model business rules and logical conditions in testing.

  Therefore, Option A is the correct answer.