Free ISQI-CT-AI Exam Dumps Preview

The syllabus explains that supervised learning requires correctly labeled data so the algorithm can learn the relationship between input features and output labels:

"In supervised learning, the algorithm creates the ML model from labeled data during the training phase.
The labeled data is used to infer the relationship between the input data and output labels." (Reference: ISTQB CT-AI Syllabus v1.0, Section 3.1.1)

The ISTQB CT-AI syllabus (Section 3.2 - Model Evaluation) specifies the correct usage of training, validation, and test data sets. It emphasizes that the test data set must be representative of the real operational data and must be equivalent in distribution to the training and validation data sets, ensuring a fair and unbiased evaluation. Option D precisely matches this requirement.

Option A contradicts the syllabus because validation and test data sets serve different purposes: validation is used for model tuning, while the test data set is used for final evaluation. Combining them undermines the reliability of results.
Option B is incorrect because even with limited data, the syllabus recommends maintaining a test data set or using techniques such as cross-validation rather than eliminating testing.
Option C is incorrect because an equal distribution between training, validation, and test data sets is not recommended; typically, the training data set is significantly larger (for example, 70-80%).

Thus, Option D is the only statement aligned with the syllabus guidance.

The ISTQB CT-AI syllabus defines flexibility and adaptability in Section 2.1 - Flexibility and Adaptability.
Flexibility is described as the ability of a system to operate in situations not explicitly covered by its original requirements, whereas adaptability refers to how easily the system can be modified to handle new environments or operating conditions. The syllabus emphasizes that both properties are especially important for self-learning AI-based systems, which may need to adjust their behavior autonomously as conditions change.

It further explains that such systems must be capable of determining when and how to adapt their behavior in evolving or previously unknown situations. This directly aligns with Option A, which highlights the importance of both adaptability and flexibility when a system needs to change its behavior and determine those changes on its own.

Option B is incorrect because it reverses the definitions: operating in unspecified situations relates to flexibility, not adaptability.
Option C is incorrect because self-learning AI-based systems are not classified as either adaptable-only or flexible-only; they typically require both characteristics. Option D is incorrect
because ease of reprogramming corresponds to adaptability, not flexibility.

Thus, Option A correctly reflects the syllabus definitions.

The ISTQB CT-AI syllabus identifies accuracy of data and labels as a key data quality aspect. Incorrectly labeled data directly indicates a problem with labeling accuracy, as the labels do not correctly represent the objects in the data. Such inaccuracies lead to misleading training signals and degrade the model's learning and predictive performance.

Security issues relate to unauthorized access or data breaches, not labeling correctness. Privacy issues concern the handling of personal or sensitive data and are unrelated to labeling errors. Bias issues involve systematic skew or unfair representation in the data, rather than labels being factually wrong.

Therefore, accuracy issues (Option B) are the most likely reason for the incorrectly labeled data.

The ISTQB CT-AI syllabus explains machine learning performance metrics in Section 3.2 - Evaluating ML
Models. For clustering, which is an unsupervised learning approach, the syllabus lists metrics such as intra-cluster distance, inter-cluster distance, and cohesion measures. Intra-cluster metrics evaluate how close data points are within the same cluster, which directly corresponds to Option A.

Option B is incorrect because R-squared is a regression metric used to measure goodness of fit and is not related to classification performance or ROC curves.
Option C is incorrect because the silhouette coefficient is a clustering metric that measures cohesion and separation between clusters, not regression model fit.
Option D is incorrect because receiver operating characteristic curves are used to evaluate binary or multiclass classification models, not clustering models.

Thus, Option A is the only statement that accurately reflects the syllabus.

Section 1.6.3 - Reinforcement Learning of the ISTQB CT-AI syllabus states that reinforcement learning (RL) is based on an agent interacting with an environment, performing actions, and receiving rewards or penalties. The core concept is the reward function, which guides the agent's learning process. The syllabus emphasizes that training in RL is driven by rewards, and the agent aims to maximize cumulative reward over time. Therefore, Option C correctly describes reinforcement learning: the agent's training is based on a reward function that rewards successful actions.

Option A is incorrect because reinforcement learning does not rely on labeled data; that is characteristic of
supervised learning.
Option B is incorrect because reinforcement learning fundamentally requires interaction with the environment.
Option D is incorrect because the reward function is defined externally (for example, by designers or domain experts); the agent learns a policy to maximize reward, not the
reward function itself.

Thus, Option C is the only statement consistent with reinforcement learning as defined in the syllabus.

The question asks which test is least likely to be performed during the ML model testing phase. Let's consider each option:

Testing the accuracy of the classification model (A): Accuracy testing is a fundamental part of the ML model testing phase. It ensures that the model correctly classifies the data as intended and meets the required performance metrics.

Testing the API of the service powered by the ML model (B): Testing the API is crucial, especially if the ML model is deployed as part of a service. This ensures that the service integrates well with other systems and that the API performs as expected.

Testing the speed of the training of the model (C): This is least likely to be part of the ML model testing phase. The speed of training is more relevant during the development phase when optimizing and tuning the model. During testing, the focus is more on the model's performance and behavior rather than how quickly it was trained.

Testing the speed of the prediction by the model (D): Testing the speed of prediction is important to ensure that the model meets performance requirements in a production environment, especially for real-time applications.

ISTQB CT-AI Syllabus Section 3.2 on ML Workflow and Section 5 on ML Functional Performance Metrics discuss the focus of testing during the model testing phase, which includes accuracy and prediction speed but not the training speed.

The ISTQB CT-AI syllabus defines data acquisition as the activity of obtaining data relevant to the business problem addressed by an ML model. This activity typically consists of three steps: identifying relevant data sources, gathering the data, and labeling the data for use in supervised learning.

The other options describe activities that belong to different phases of the ML workflow. Cleaning and transforming are part of data preprocessing, feature-related activities are part of feature engineering, and building or deploying relates to later stages of system development.

Therefore, Identifying, gathering, labeling (Option C) is the correct answer.

The low prevalence rate (2%) indicates a highly imbalanced dataset, where cancer cases represent a very small minority of the population. In such scenarios, accuracy alone is misleading, as a model could achieve high accuracy by predominantly predicting the majority class (normal patients).

Recall (sensitivity) is critical in medical diagnosis because it measures the proportion of actual cancer cases correctly identified. Missing cancer cases (false negatives) can have severe consequences, so recall must be maximized.

Precision is also important because it measures how many predicted cancer cases are truly cancer cases.
High precision reduces false positives, avoiding unnecessary anxiety and invasive follow-up procedures for healthy patients.

When both false negatives and false positives have significant impact, as in cancer detection, the ISTQB CT-AI syllabus emphasizes the need to maximize both recall and precision, rather than relying on accuracy or specificity alone.

Therefore, Option C (Maximize recall and precision) is the correct answer.

The scenario describes a model that shows good accuracy in isolation but performs poorly in real-world operation, particularly in detecting true cancer cases. This strongly indicates a generalization problem.

A lack of similarity between the training and testing data means that the training data was not sufficiently representative of real-world diagnostic data. This is a well-known ML risk that results in models performing well during development but failing in operational environments.

While input data quality (Option B) is important, the issue described is not primarily about corrupted or invalid data but about representativeness.

Choosing inappropriate functional performance metrics (Option C) could hide problems, but it would not by itself cause poor real-world detection if the data were representative.

Non-functional requirements testing (Option D) is important but does not directly explain the inability to
detect real cancer cases.

The ISTQB CT-AI syllabus (Section 4.2 - Training, Validation, and Test Datasets) emphasizes that a lack of representative data across training, validation, and operational environments is a major cause of poor
real-world ML performance.

Therefore, Option A is the most likely deficiency discovered by the test expert.

Section 1.7 - Neural Networks of the ISTQB CT-AI syllabus explains that neural networks consist of neurons connected by weighted links. During training, learning occurs by adjusting the weights on these connections. This is the essence of gradient descent and backpropagation. Option B correctly states this behavior: only the weights are modified, not the activation functions, the number of neurons, or the architectural structure.

Option A is incorrect because a neuron's bias is not determined by the activations of neurons in the previous layer; it is an independent trainable parameter added to the weighted sum of inputs. Option C is
incorrect because the syllabus states that a single-layer perceptron is a valid type of neural network, although it is limited to linearly separable problems. Option D is incorrect because there is no requirement that the number of input neurons be greater than or equal to the number of output neurons. Input neurons correspond to the number of features, while output neurons correspond to tasks or classes.

Therefore, Option B precisely reflects the syllabus definition of what changes during neural network training.

The ISTQB CT-AI syllabus explains in Section 3.2 - Functional Performance Criteria of ML Models that accuracy becomes unreliable when class imbalance exists. In fraud detection, more than 99% of transactions are non-fraudulent, meaning the data set is extremely imbalanced. Because accuracy counts all correctly classified non-fraudulent transactions, it can appear artificially high even when the fraud detection performance is poor. Therefore, accuracy is not suitable for evaluating fraud detection systems.

The syllabus further explains that sensitivity (recall) measures the proportion of correctly identified fraudulent transactions. This metric is important because missed fraudulent events can lead to significant financial loss. However, the client also requires that legitimate transactions be correctly identified, meaning false positives must be minimized to maintain customer satisfaction.

The F1 score, defined as the harmonic mean of precision and recall, balances both aspects: precision limits false alarms affecting legitimate customers, while recall ensures that fraudulent transactions are detected. Section 3.2 emphasizes that when both false positives and false negatives have serious consequences and the data is highly imbalanced, the F1 value is the most appropriate metric.

Thus, Option C is the correct choice.

The ISTQB CT-AI syllabus defines AI-based autonomous systems as systems that can perceive and respond to their environment and make decisions or take actions without direct human intervention.
Autonomy implies that the system can adapt its behavior based on environmental inputs.

Option D correctly describes such a system, as the ability to respond to the environment independently is a core characteristic of AI-based autonomous functionality.

Option A is incorrect because decision-making is entirely performed by humans.
Option B is incorrect because a system that uses no software cannot implement AI-based autonomy.
Option C is incorrect because Selenium is a test automation tool and does not represent an autonomous AI system.

Therefore, Option D is the correct answer.

For AI-enabled medical devices operating in life-critical contexts, the ISTQB CT-AI syllabus emphasizes that certification must focus primarily on aspects that directly impact patient safety and regulatory trust.

Safety (Aspect III) is essential to ensure that the AI system does not cause harm to patients.
Transparency (Aspect IV) is required so that medical professionals and regulators can understand, assess, and justify the AI's diagnostic decisions. Side effects (Aspect V) must be identified and controlled, as unintended or emergent behaviors can have serious consequences in medical applications.

Autonomy and maintainability are relevant quality characteristics but are secondary compared to safety-critical concerns such as preventing harm, ensuring explainability, and managing unintended effects.

Therefore, Aspects III, IV, and V are the three most required aspects for certification in this scenario.

The machine learning workflow typically proceeds iteratively, but once the objectives, business priorities, framework, and model have been agreed upon, the next step is data preparation. The ISTQB CT-AI syllabus explains that data preparation is required to provide suitable input data for training an ML algorithm and for evaluating the resulting model.

The syllabus notes that ML models cannot be trained or evaluated without properly prepared data and that data preparation is therefore a prerequisite for subsequent steps such as training, tuning, and evaluation.
This includes collecting, cleaning, transforming, and splitting the data for training and testing purposes.

This directly supports Option A as the correct next step in the machine learning workflow.

Back-to-back testing is a technique in which the same test cases and input data are executed against multiple implementations that are intended to provide equivalent behavior, and their outputs are compared to identify inconsistencies.

Option A describes comparing two implementations of the same type of model (neural networks) built on different platforms using identical data, which is a typical back-to-back testing scenario.
Option B describes comparing a custom implementation with a standard reference implementation using the same data, which is also a common use of back-to-back testing to validate correctness.

Option C describes comparing a neural network model with a decision tree model using the same data.
Because these models are fundamentally different in structure and behavior, they are not expected to produce equivalent outputs, making this the least appropriate scenario for back-to-back testing.

Option D compares the same model using different datasets. This represents robustness or sensitivity testing rather than back-to-back testing, but it still does not involve comparing different model types.

Therefore, Option C is the correct answer, as it least describes a back-to-back testing situation.

According to Section 4.6 --- AI Behaviour Testing (AIB Testing) of the ISTQB CT-AI syllabus, AIB testing is used to evaluate changes in the functional behavior of self-learning systems. The core principle is comparing pre-change and post-change model behavior using the same test inputs so that any differences in outputs can be attributed to the model's learning rather than to changes in the input data.
This directly corresponds to Option C.

Option A is incorrect because the absence of a test oracle does not justify generating new test cases; AIB testing relies on reusing identical inputs to detect behavioral drift.
Option B is incorrect because using different inputs prevents meaningful comparison of behavior changes.
Option D is incorrect because comparing a self-learning system with an unrelated non-self-learning system does not allow evaluation of how the same model's behavior has evolved over time.

Thus, Option C accurately represents the correct application of AIB testing.

Detecting biases in the ML pipeline involves various tests to ensure fairness and accuracy throughout the
ML process.

Testing the distribution shift in the training data for inappropriate bias (A): This involves checking if there is any shift in the data distribution that could lead to bias in the model. It is an important test but not the most direct method for detecting biases.

Test the model during model evaluation for data bias (B): This is a critical stage where the model is evaluated to detect any biases in the data it was trained on. It directly addresses potential data biases in the model.

Testing the data pipeline for any sources for algorithmic bias (C): This test is crucial as it helps identify biases that may originate from the data processing and transformation stages within the pipeline. Detecting sources of algorithmic bias ensures that the model does not inherit biases from these processes.

Check the input test data for potential sample bias (D): While this is an important step, it focuses more on the input data and less on the overall data pipeline. Hence, the most likely useful test to help detect different kinds of biases in the ML pipeline isB. Test the model during model evaluation for data bias .