100+ Free ISTQB CT-AI Practice Questions

Pass your ISTQB Certified Tester — AI Testing Specialist (CT-AI v1.0) exam on the first try — instant access, no signup required.

✓ No registration✓ No credit card✓ No hidden fees✓ Start practicing immediately

~70-80% Pass Rate

100+ Questions

100% Free

1 / 10

Question 1

Score: 0/0

Which ISO standard extends ISO/IEC 25010 with quality characteristics specific to AI-based systems?

ISO/IEC 25010

ISO/IEC 25012

ISO/IEC 25059

ISO/IEC 27001

to track

2026 Statistics

Key Facts: ISTQB CT-AI Exam

Exam Questions

ISTQB

26/40

Passing Score

65%

60 min

Exam Duration

75 min non-native

$200-$249

Exam Fee

ISTQB Specialist

Lifetime

Cert Valid

No renewal

CTFL

Prerequisite

Foundation Level required

The ISTQB CT-AI v1.0 exam has 40 multiple-choice questions in 60 minutes (75 min for non-native English speakers) with a 65% passing score (26/40). Major chapters: Introduction to AI, Quality Characteristics for AI-Based Systems, Machine Learning, ML Data, ML Functional Performance Metrics, Neural Networks, Methodologies for Testing AI Systems, AI for Testing. Exam fee is $200-$249 USD. Requires CTFL Foundation. Certification is valid for life.

Sample ISTQB CT-AI Practice Questions

Try these sample questions to test your ISTQB CT-AI exam readiness. Each question includes a detailed explanation. Start the interactive quiz above for the full 100+ question experience with AI tutoring.

1Which ISO standard extends ISO/IEC 25010 with quality characteristics specific to AI-based systems?

A.ISO/IEC 25010

B.ISO/IEC 25012

C.ISO/IEC 25059

D.ISO/IEC 27001

Explanation: ISO/IEC 25059 is the AI quality model that extends ISO/IEC 25010 by adding AI-specific characteristics: flexibility, autonomy, evolution, and transparency. ISO/IEC 25010 defines the general software product quality model (functional suitability, performance efficiency, reliability, usability, security, maintainability, portability, compatibility). ISO/IEC 25012 covers data quality, and ISO/IEC 27001 is an information security management standard.

2Which of the following is NOT one of the AI-specific quality characteristics defined by ISO/IEC 25059?

A.Flexibility

B.Autonomy

C.Maintainability

D.Transparency

Explanation: Maintainability is a general software quality characteristic from ISO/IEC 25010, not an AI-specific characteristic added by ISO/IEC 25059. The AI-specific characteristics are flexibility (the ability to deal with situations beyond initial requirements), autonomy (operation without human intervention), evolution (continued operation over time as data changes), and transparency (explainability of decisions).

3What term describes the phenomenon where a capability stops being considered AI once it becomes commonplace?

A.AI winter

B.AI effect

C.AI singularity

D.AI saturation

Explanation: The AI effect describes how once an AI capability becomes widely used and understood (such as optical character recognition or chess engines), it tends to no longer be considered 'real' AI. AI winter refers to historical periods of reduced AI funding and interest. The technological singularity refers to a hypothetical point when AI surpasses human intelligence. AI saturation is not a standard term.

4Which type of machine learning trains a model using labeled input-output pairs?

A.Unsupervised learning

B.Supervised learning

C.Reinforcement learning

D.Self-supervised learning

Explanation: Supervised learning uses datasets where each input has a known correct output (label). Examples include classification (spam vs not spam) and regression (predicting house prices). Unsupervised learning works with unlabeled data (e.g., clustering). Reinforcement learning trains via rewards and penalties based on actions taken in an environment. Self-supervised learning generates labels from the data itself, often used in foundation model pretraining.

5Which ML task assigns inputs to one of a fixed set of discrete categories?

A.Regression

B.Clustering

C.Classification

D.Dimensionality reduction

Explanation: Classification assigns inputs to one of a predefined set of discrete classes (e.g., spam/not-spam, cat/dog/bird). Regression predicts a continuous numeric value (e.g., temperature, price). Clustering groups similar items without predefined labels. Dimensionality reduction (e.g., PCA, t-SNE) reduces feature count while preserving structure.

6In a typical ML workflow, what is the purpose of a validation set (separate from train and test sets)?

A.To produce the final published accuracy figure

B.To tune hyperparameters and select the best model without leaking the test set

C.To train the model

D.To measure data drift in production

Explanation: The validation set is used during development to tune hyperparameters, select architectures, and choose between candidate models. The test set is held back and used only once for the final, unbiased performance estimate; using the test set for tuning would cause information leakage and inflate the reported metric. The training set fits the model parameters. Drift detection happens in production with monitoring, not the validation set.

7What is k-fold cross-validation primarily used for?

A.Reducing training time

B.Producing a more reliable performance estimate from limited data

C.Increasing the size of the training set

D.Eliminating class imbalance

Explanation: K-fold cross-validation partitions the data into k folds and trains/evaluates k times, each time using a different fold as the validation set. Averaging the k results produces a more stable, lower-variance performance estimate, particularly valuable when data is limited. It does not reduce training time (it actually increases compute), does not enlarge the dataset, and does not by itself fix class imbalance.

8What is data leakage in a machine learning context?

A.Sensitive data being exposed to attackers

B.Information from outside the training set inadvertently influencing the model

C.Loss of training data due to disk failure

D.Memory leaks during model training

Explanation: Data leakage occurs when information that would not be available at prediction time leaks into training — for example, a feature derived from the target, or test-set statistics being used in feature engineering on the training set. It causes inflated training/validation metrics that collapse in production. The other options describe security breaches or infrastructure issues, not the ML concept.

9Which sampling technique preserves the class distribution between train and test sets?

A.Random sampling

B.Stratified sampling

C.Convenience sampling

D.Bootstrap sampling

Explanation: Stratified sampling ensures each subset (train/test/validation) contains the same proportion of each class as the original dataset. This is critical for imbalanced datasets — random sampling could produce a test set with very few minority-class examples, distorting metrics. Bootstrap sampling draws with replacement and is used for ensembles like random forests, not strictly for class preservation.

10What is the typical first phase of the ML workflow?

A.Hyperparameter tuning

B.Data acquisition

C.Model deployment

D.Drift monitoring

Explanation: The ML workflow typically begins with data acquisition (collecting and assembling the raw dataset), followed by exploratory data analysis (EDA), feature engineering, model training, validation, testing, deployment, and finally production monitoring. Hyperparameter tuning happens after a baseline model exists; deployment and drift monitoring are late-stage activities.

About the ISTQB CT-AI Exam

The ISTQB Certified Tester AI Testing (CT-AI v1.0) is an ISTQB Specialist certification that validates skills to test AI-based systems and apply AI techniques in testing. The syllabus covers AI quality characteristics (ISO/IEC 25059), the ML lifecycle, data quality and bias, model performance metrics (precision, recall, F1, ROC-AUC), neural networks, adversarial and metamorphic testing, MLOps testing, and AI-driven test activities. Requires the ISTQB Foundation Level (CTFL) as a prerequisite.

Questions

40 scored questions

Time Limit

60 minutes

Passing Score

65% (26/40)

Exam Fee

$200-$249 USD (ISTQB / Pearson VUE)

ISTQB CT-AI Exam Content Outline

Introduction to AI

AI definition, narrow vs general vs super AI, AI effect, technological singularity, AI-based vs conventional systems

15%

Quality Characteristics for AI-Based Systems

ISO/IEC 25059 quality model — flexibility, autonomy, evolution, transparency, plus ISO 25010 functional suitability, performance efficiency, compatibility, usability, reliability, security, maintainability, portability

15%

Machine Learning Overview

Supervised vs unsupervised vs reinforcement learning, classification vs regression vs clustering, ML workflow, train/validation/test split, k-fold cross validation

15%

ML — Data

Data acquisition, exploratory data analysis (EDA), feature engineering, data quality (completeness, consistency, accuracy), missing data, outliers, duplicates, label noise, class imbalance, data leakage, sampling bias

15%

ML Functional Performance Metrics

Confusion matrix (TP/TN/FP/FN), accuracy, precision, recall, F1, F-beta, MCC, ROC-AUC, PR-AUC; regression metrics (MAE, MSE, RMSE, MAPE, R²); selecting metrics for use cases

Neural Networks and Testing

Neural network basics, deep learning, coverage measures for neural networks (neuron coverage, sign coverage)

20%

Methodologies for Testing AI Systems

Adversarial examples (FGSM, PGD), metamorphic testing, A/B testing for models, shadow and canary deployment, test pyramid for ML, bias and fairness testing (demographic parity, equalized odds), explainability (SHAP, LIME), monitoring (data drift, concept drift, KS test, PSI)

10%

Using AI for Testing

AI-driven test case generation, defect prediction, visual testing, log analysis, test optimization with ML, MLflow/W&B for tracking

How to Pass the ISTQB CT-AI Exam

What You Need to Know

Passing score: 65% (26/40)
Exam length: 40 questions
Time limit: 60 minutes
Exam fee: $200-$249 USD

Keys to Passing

Complete 500+ practice questions
Score 80%+ consistently before scheduling
Focus on highest-weighted sections
Use our AI tutor for tough concepts

ISTQB CT-AI Study Tips from Top Performers

1Memorize the ISO/IEC 25059 AI-specific quality characteristics: flexibility, autonomy, evolution, transparency

2Master the confusion matrix and be able to compute precision, recall, F1, and accuracy by hand

3Understand when to prefer recall over precision (medical screening) vs precision over recall (spam, fraud alerts)

4Study metamorphic testing relations — they appear in K3/K4 application questions

5Know the difference between data drift, concept drift, and prediction drift, and detection methods (KS test, PSI)

6Understand bias categories: sampling, measurement, label, historical, algorithmic, automation, confirmation

7Be comfortable with the ML workflow: data acquisition → EDA → feature engineering → train → validate → test → deploy → monitor

8Complete all 100 practice questions and review every wrong-answer explanation

Frequently Asked Questions

What is the ISTQB CT-AI exam?

The ISTQB Certified Tester AI Testing (CT-AI v1.0) is a Specialist-level ISTQB certification covering how to test AI-based systems and how to use AI in testing. It addresses ML quality characteristics from ISO/IEC 25059, the ML lifecycle, data quality, performance metrics, neural networks, adversarial and metamorphic testing, and MLOps. The Foundation Level (CTFL) is a prerequisite.

How many questions are on the CT-AI exam and what is the passing score?

The CT-AI exam has 40 multiple-choice questions to be completed in 60 minutes (75 minutes for non-native English speakers). The passing score is 65%, which equals 26 of 40 correct answers. Some scoring models weight harder K3/K4 questions slightly higher, but 65% remains the threshold.

What does CT-AI cost in 2026?

ISTQB Specialist exams including CT-AI typically cost between $200 and $249 USD in the United States via ASTQB and iSQI. Pricing varies by national board and whether you book the exam alone or bundle with accredited training. Exact pricing is published on the istqb.org and astqb.org certification pages.

Do I need CTFL before taking CT-AI?

Yes — the ISTQB Foundation Level (CTFL) is a formal prerequisite for the CT-AI Specialist exam. ASTQB and other national boards verify your CTFL credential before allowing you to register. There are no other formal experience requirements, but practical exposure to ML projects helps significantly.

What ML metrics does CT-AI emphasize?

Confusion matrix terminology (TP, TN, FP, FN) and the metrics derived from it: accuracy, precision, recall, F1 (and F-beta), MCC, ROC-AUC, and PR-AUC. For regression, MAE, MSE, RMSE, MAPE, and R². The syllabus also stresses that no single metric fits every use case — for example, recall matters for cancer detection while precision matters for spam filtering.

What is metamorphic testing for AI systems?

Metamorphic testing uses metamorphic relations (MRs) to verify model behavior when an oracle is hard to define. Example MRs: small input perturbations should not change classification (MR1); relabeling training data and re-training should produce predictions consistent with the new labels (MR2). It is heavily emphasized in the CT-AI syllabus because exact ground-truth oracles often don't exist for ML models.

How long should I study for CT-AI?

Plan 30-50 hours over 4-6 weeks if you are an experienced tester new to ML, or 20-30 hours if you already work with ML systems. The ISTQB recommends accredited training (about 25 instructional hours). Read the CT-AI v1.0 syllabus, complete 100+ practice questions, and aim for 80%+ before booking.