Docsity
Docsity
Prepare for your exams
Prepare for your exams

Study with the several resources on Docsity

Earn points to download
Earn points to download

Earn points by helping other students or get them with a premium plan

Guidelines and tips
Guidelines and tips
Sell on Docsity
Sell on Docsity
Docsity AI
Docsity AI
Log inSign up

Prepare for your exams

Study with the several resources on Docsity

Find documents

Prepare for your exams with the study notes shared by other students like you on Docsity

Search for your university

Find the specific documents for your university's exams

Docsity AINEW

Summarize your documents, ask them questions, convert them into quizzes and concept maps

Explore questions

Clear up your doubts by reading the answers to questions asked by your fellow students

Earn points to download

Earn points by helping other students or get them with a premium plan

Share documents
download point icon20 Points

For each uploaded document

Answer questions
download point icon5 Points

For each given answer (max 1 per day)

All the ways to get free points
Get points immediately

Choose a premium plan with all the points you need

Study Opportunities

Choose your next study program

Get in touch with the best universities in the world. Search through thousands of universities and official partners

Community

Ask the community

Ask the community for help and clear up your study doubts

Free resources

Our save-the-student-ebooks!

Download our free guides on studying techniques, anxiety management strategies, and thesis advice from Docsity tutors

British Columbia Artificial Intelligence Practitioner Certification Exam Practice Questi, Exams of Artificial Intelligence

American Samoa Community CollegeArtificial Intelligence

British Columbia Artificial Intelligence Practitioner Certification Exam Practice Questions And Correct Answers (Verified Answers) Plus Rationale 2026 Q&A| Instant Download Pdf

Typology: Exams

2025/2026

Available from 04/16/2026

Theexamwhisperer
Theexamwhisperer 🇺🇸

3

(9)

24K documents

1 / 25

Toggle sidebar

This page cannot be seen from the preview

Don't miss anything!

bg1
British Columbia Artificial Intelligence
Practitioner Certification Exam Practice
Questions And Correct Answers
(Verified Answers) Plus Rationale 2026
Q&A| Instant Download Pdf
1. A company implements an AI-driven recommendation engine that
continually adjusts product suggestions based on real-time user
interaction data. Which of the following best describes the system’s
learning paradigm?
A. Supervised learning
B. Reinforcement learning
C. Unsupervised learning
D. Transfer learning
Reinforcement learning is correct because the system iteratively
adjusts its behavior in response to user feedback to optimize
outcomes over time.
2. An AI team is designing a model to classify customer support tickets
into categories (billing, technical issue, account management). Which
evaluation metric is most appropriate when false negatives have a
higher business cost than false positives?
A. Accuracy
B. F1 Score
C. Recall
D. Precision
Recall is correct because it prioritizes minimizing false negatives,
pf3
pf4
pf5
pf8
pf9
pfa
pfd
pfe
pff
pf12
pf13
pf14
pf15
pf16
pf17
pf18
pf19

Partial preview of the text

Download British Columbia Artificial Intelligence Practitioner Certification Exam Practice Questi and more Exams Artificial Intelligence in PDF only on Docsity!

British Columbia Artificial Intelligence

Practitioner Certification Exam Practice

Questions And Correct Answers

(Verified Answers) Plus Rationale 2026

Q&A| Instant Download Pdf

  1. A company implements an AI-driven recommendation engine that continually adjusts product suggestions based on real-time user interaction data. Which of the following best describes the system’s learning paradigm? A. Supervised learning B. Reinforcement learning C. Unsupervised learning D. Transfer learning Reinforcement learning is correct because the system iteratively adjusts its behavior in response to user feedback to optimize outcomes over time.
  2. An AI team is designing a model to classify customer support tickets into categories (billing, technical issue, account management). Which evaluation metric is most appropriate when false negatives have a higher business cost than false positives? A. Accuracy B. F1 Score C. Recall D. Precision Recall is correct because it prioritizes minimizing false negatives,

which is critical when missing a category has significant consequences.

  1. During an AI project’s data preparation phase, the team discovers that several features have missing values. Which method is generally considered least appropriate if the data is not missing completely at random? A. Mean imputation B. Dropping rows with missing values C. Model-based imputation D. Using algorithms that handle missing values internally Dropping rows is correct because it can bias results when missingness is systemic rather than random.
  2. In the context of convolutional neural networks (CNNs), what is the primary role of a pooling layer? A. Apply activation functions B. Normalize input data C. Reduce spatial dimensions D. Increase model complexity Pooling reduces spatial dimensions to lower computational load and help the network focus on salient features.
  3. A data scientist uses principal component analysis (PCA) before training a classification model. What is the main purpose of PCA in this scenario? A. Enhancing class separability B. Creating more features C. Reducing dimensionality D. Balancing the dataset PCA is used to reduce dimensionality by transforming the original features into principal components that capture the most variance.
  4. Which of the following best illustrates a situation where using gradient descent would be necessary? A. Computing descriptive statistics B. Validating a decision tree model C. Minimizing a loss function for a neural network D. Encoding categorical variables

Overfitting occurs when the model learns training data too well, failing to generalize to validation data.

  1. What principle requires that data used for AI model training is collected and processed only with explicit user consent, where applicable? A. Data minimization B. Model interpretability C. Explainability D. Privacy and consent Privacy and consent principles mandate that individuals must agree to the use of their data for AI training and processing.
  2. Which of the following best defines transfer learning? A. Training from scratch on large datasets B. Using unsupervised clustering to label data C. Leveraging a pretrained model for a new, related task D. Reducing overfitting using dropout Transfer learning uses an existing model’s learned features to accelerate and improve training for a related but different task.
  3. A sensitivity analysis shows that a trained model’s predictions vary greatly when a particular feature value changes slightly. What does this suggest about that feature? A. It is irrelevant B. It has low variance C. It significantly influences the model’s prediction D. It should be removed Significant prediction variation with small feature changes indicates the feature strongly influences model output.
  4. Which of the following strategies is most effective for addressing class imbalance in a binary classification dataset? A. Increasing learning rate B. Oversampling the minority class C. Reducing dataset size D. Eliminating cross-validation Oversampling the minority class helps balance class representation, improving the classifier’s ability to detect rare cases.
  1. In regression analysis, which assumption relates to the expectation that residuals exhibit constant variance across all levels of the independent variable? A. Linearity B. Independence C. Homoscedasticity D. Normality Homoscedasticity refers to residuals having constant variance instead of varying with fitted values.
  2. A developer implements batch normalization in a deep learning model primarily to: A. Increase model depth B. Stabilize and speed up training C. Reduce input size D. Eliminate activation functions Batch normalization reduces internal covariate shift, leading to more stable and faster model training.
  3. Which scenario best exemplifies the use of AI for predictive maintenance? A. Chatbot handling customer queries B. Image classification of product defects C. Forecasting machine breakdowns from sensor data D. Translating documentation Predictive maintenance uses sensor data to anticipate equipment failures before they occur.
  4. In the context of AI deployment, what is the purpose of A/B testing? A. Data labeling automation B. Model compression C. Comparing performance of two system versions D. Clustering similar data points A/B testing evaluates which version of a system or feature performs better in real user scenarios.
  5. What does the term “explainability” in AI refer to? A. Compression of neural networks

D. K-Means clustering Recurrent networks process sequences by maintaining internal state based on previous inputs.

  1. In risk assessment for an AI initiative, which factor most directly concerns the probability of erroneous model predictions causing harm? A. Infrastructure cost B. Dataset size C. Model error rates in critical categories D. Hyperparameter choices Model error rates in contexts where incorrect outputs have serious consequences directly inform risk severity.
  2. Which of the following is an ethical issue specific to autonomous decision-making systems? A. Database normalization B. Scaling cloud resources C. Algorithmic accountability D. Feature standardization Algorithmic accountability addresses who is responsible when autonomous systems make harmful or questionable decisions.
  3. When deploying an AI model on edge devices, which constraint is most critical to consider? A. Number of training epochs B. Cloud service tiers C. Computational and memory limitations D. Dataset diversity Edge deployments require models to be optimized for limited processor and memory capacities.
  4. Which approach to handling high-cardinality categorical variables reduces dimensionality effectively? A. One-hot encoding B. Mean imputation C. Target encoding D. Z-score normalization

Target encoding represents categories by aggregated target statistics, reducing dimensionality compared to one-hot encoding.

  1. An organization wants to ensure continuous monitoring of an AI model in production. What practice supports detecting performance degradation over time? A. Manual chart review quarterly B. Automated performance tracking with alerts C. Disabling validation checks D. Removing ground truth labels Automated tracking with alerts ensures timely detection and correction of degrading performance.
  2. Which activation function is known for mitigating the vanishing gradient problem in deep neural networks? A. Sigmoid B. Tanh C. ReLU D. Softmax ReLU maintains gradients for positive inputs, reducing vanishing gradient issues compared with sigmoid or tanh.
  3. In cluster analysis, which method determines membership by minimizing distance to multiple centroids? A. Hierarchical clustering B. K-Means clustering C. Linear discriminant analysis D. Naive Bayes classifier K-Means partitions data around centroids by minimizing within-cluster distances.
  4. When considering the fairness of an AI hiring tool, which concept ensures that decisions do not disproportionately disadvantage a protected group? A. Overfitting B. Market segmentation C. Demographic parity D. Feature scaling
  1. What is a key advantage of using ensemble methods like random forests over a single decision tree? A. Higher training speed B. Simplified interpretability C. Reduced overfitting and improved generalization D. Lower memory usage Ensembles average diverse models’ predictions, typically reducing variance and improving performance.
  2. In speech-to-text systems, which technology primarily converts acoustic signals into linguistic units? A. Generative adversarial networks B. Acoustic modeling C. Convolutional layers D. Tokenization Acoustic modeling translates audio features into probable phonemes and words.
  3. A model deployed in production begins to encounter data that differ significantly from training data. This phenomenon is known as: A. Underfitting B. Regularization C. Data drift D. Hyperparameter tuning Data drift describes changes in input data distributions after model deployment.
  4. In AI software engineering, what practice ensures that model code changes don’t unintentionally break existing functionality? A. Manual testing only B. Ignoring unit tests C. Continuous integration with automated tests D. Disabling version control Continuous integration with automated testing catches regressions early during development.
  5. Which of the following best describes a confusion matrix in classification tasks? A. A graph of model loss over time

B. A plot of precision vs recall C. A table showing true/false positives and negatives D. A matrix of feature correlations A confusion matrix presents classification counts for predicted vs actual classes.

  1. In optimization algorithms, what is the role of a learning rate? A. It increases dataset size B. It controls step size during parameter updates C. It alters activation functions D. It splits data into batches The learning rate dictates how much model parameters are adjusted each iteration.
  2. When building an AI model that must be transparent to regulators, which attribute is most important? A. Large parameter count B. High complexity C. Interpretability D. Random initialization Interpretability ensures that stakeholders can understand how the model reaches decisions.
  3. Which loss function is most appropriate for multi-class classification with mutually exclusive labels? A. Mean squared error B. Hinge loss C. Categorical cross-entropy D. Cosine similarity Categorical cross-entropy measures prediction error across multi-class probabilities effectively.
  4. What is the primary purpose of regularization techniques like L and L2 in model training? A. Increase model complexity B. Remove all features C. Prevent overfitting D. Expand datasets
  1. When preparing data for supervised learning, which process assigns labelled outcomes to raw inputs? A. Feature scaling B. Data annotation C. Grid search D. Model compression Data annotation labels raw inputs, creating labeled pairs needed for supervised training.
  2. In ethical AI, what principle demands that individuals be able to contest automated decisions affecting them? A. Model accuracy B. Computational efficiency C. Right to explanation and appeal D. Hyperparameter optimization The right to explanation and appeal ensures individuals can challenge decisions made by automated systems.
  3. A company wants to use AI to detect fraudulent transactions in real time. Which type of learning is most appropriate for this task? A. Supervised learning with historical labeled transactions B. Unsupervised clustering of customer segments C. Reinforcement learning for reward optimization D. Semi-supervised learning combining labeled and unlabeled data Semi-supervised learning is effective here because labeled fraud examples are limited, so combining them with abundant unlabeled data improves detection performance.
  4. In AI model evaluation, what is the purpose of stratified sampling during cross-validation? A. To reduce computation time B. To maintain class distribution in each fold C. To remove outliers from the dataset D. To randomly shuffle features Stratified sampling ensures that each fold in cross-validation

preserves the original class distribution, preventing bias in performance evaluation.

  1. Which scenario best exemplifies unsupervised learning? A. Predicting house prices based on features B. Classifying emails as spam or not spam C. Grouping customers into segments without pre-defined labels D. Forecasting sales using past revenue data Unsupervised learning discovers patterns in data without labeled outputs, such as clustering customers based on behavior.
  2. Which regularization technique is particularly effective in driving model sparsity by eliminating less important features? A. L2 (Ridge) B. L1 (Lasso) C. Dropout D. Early stopping L1 regularization (Lasso) can shrink some coefficients to zero, effectively performing feature selection and reducing model complexity.
  3. Which type of AI system relies on explicitly programmed rules rather than learning from data? A. Deep neural networks B. Decision trees C. Expert systems D. Reinforcement learning agents Expert systems encode domain knowledge as rules to make inferences, rather than learning patterns from data.
  4. In reinforcement learning, what does the term “exploration” refer to? A. Repeating known actions for reward B. Trying new actions to discover potential rewards C. Optimizing hyperparameters D. Reducing model complexity Exploration involves taking novel actions to gather information about the environment, balancing with exploitation of known rewards.

A. Standard neural networks B. Decision trees C. Bayesian models D. K-Means clustering Bayesian models incorporate probability distributions, allowing them to quantify uncertainty in predictions.

  1. In natural language processing, what is the primary purpose of tokenization? A. Reducing model size B. Splitting text into meaningful units such as words or subwords C. Removing irrelevant features D. Encoding numerical values Tokenization converts text into discrete units that models can process efficiently.
  2. A company wants to deploy a machine learning model that continues to learn from new data after deployment. Which approach is most appropriate? A. Static training only B. Batch processing once C. Online learning D. Transfer learning on unrelated data Online learning updates the model incrementally as new data arrives, allowing adaptation in real time.
  3. Which metric is best suited for evaluating a regression model’s ability to predict continuous numerical outcomes? A. Accuracy B. F1 Score C. Confusion matrix D. Root Mean Squared Error (RMSE) RMSE measures the average magnitude of prediction errors in continuous data, providing a standard metric for regression performance.
  4. What is a primary risk when deploying a model trained on historical data in a dynamic environment? A. Overfitting

B. Hyperparameter tuning C. Concept drift D. Regularization Concept drift occurs when the underlying relationships in data change over time, potentially degrading model performance.

  1. Which AI approach combines multiple weak learners to create a stronger predictive model? A. Support vector machines B. K-Means clustering C. Ensemble learning D. Neural networks Ensemble learning aggregates multiple models to improve prediction accuracy and robustness.
  2. In a neural network, which layer type is typically used to reconstruct input data in unsupervised learning tasks? A. Convolutional layer B. Fully connected layer C. Autoencoder layer D. Recurrent layer Autoencoder layers compress and reconstruct input data, useful for unsupervised representation learning.
  3. When building AI for healthcare diagnosis, why is explainability particularly critical? A. To reduce computation cost B. To ensure clinicians can trust and validate model decisions C. To increase dataset size D. To speed up model training Explainability allows healthcare professionals to understand and validate AI decisions, which is essential for patient safety.
  4. Which of the following best describes the role of a discount factor in reinforcement learning? A. Determines learning rate B. Selects features C. Weighs the importance of future rewards D. Reduces overfitting
  1. A company wants an AI system to detect both known and previously unseen malware. Which approach is most suitable? A. Pure supervised learning B. Rule-based system only C. Hybrid approach combining supervised and anomaly detection D. Transfer learning from unrelated domains A hybrid approach captures patterns of known malware while detecting anomalies that may represent new threats.
  2. In deep learning, which optimizer adapts learning rates for each parameter individually based on first and second moments of gradients? A. SGD B. RMSProp C. Adam D. Adagrad Adam optimizer adjusts learning rates adaptively using estimates of first and second moments, improving convergence.
  3. Which AI concept refers to unintended systematic errors resulting from training data or algorithm design? A. Variance B. Noise C. Bias D. Drift Bias reflects systematic errors that cause models to consistently deviate from correct predictions.
  4. In NLP, which technique captures word meaning based on surrounding context rather than fixed vectors? A. One-hot encoding B. Bag-of-words C. Contextual embeddings (e.g., BERT) D. TF-IDF Contextual embeddings produce dynamic representations for words depending on their context, improving semantic understanding.
  5. Which strategy reduces the risk of overfitting in small datasets by generating additional plausible training samples?

A. Feature scaling B. Data augmentation C. Hyperparameter tuning D. Principal component analysis Data augmentation artificially expands the dataset by applying transformations, enhancing model generalization.

  1. Which neural network architecture is specifically designed for attention-based processing of sequential data? A. Convolutional Neural Network B. Recurrent Neural Network C. Transformer D. Autoencoder Transformers use self-attention mechanisms to process sequences efficiently, allowing parallel computation and long-range dependency modeling.
  2. In AI governance, what is the primary purpose of an audit trail? A. Increase model complexity B. Reduce training time C. Document decisions, data, and processes for accountability D. Minimize feature set Audit trails record the steps and decisions in AI development and deployment, enabling accountability and regulatory compliance.
  3. Which method evaluates the importance of individual features by measuring performance degradation when a feature is permuted? A. PCA B. Correlation analysis C. Permutation feature importance D. Dropout Permutation feature importance assesses the impact of each feature by observing changes in model performance when its values are shuffled.
  4. Which problem in AI arises when a model relies on spurious correlations present in training data rather than causal relationships? A. Underfitting B. Regularization