1. A hospital deploys an AI model to detect pneumonia from chest X-rays. After deployment, doctors notice the model performs poorly on images from a new machine. What issue is most likely occurring?

• Overfitting
• Domain shift
• Underfitting
• Data leakage

2. A bank uses AI to predict loan defaults. Regulators demand transparency in how predictions are made. Which technique should the bank use?

• Gradient descent
• PCA
• Bagging
• SHAP values

3. An e-commerce site uses AI to recommend products. Customers complain recommendations are repetitive. What technique can improve diversity?

• Collaborative filtering with diversity constraints
• Logistic regression
• Naive Bayes
• Feature scaling

4. A self-driving car must decide whether to stop or continue at a yellow light. Which AI paradigm best fits this decision-making process?

• Supervised learning
• Reinforcement learning
• Unsupervised learning
• Deep learning

5. A fraud detection system flags too many legitimate transactions as fraud. Which metric should be prioritized to reduce false alarms?

• Precision
• Recall
• Accuracy
• F1-score

6. A company trains a chatbot but finds it gives offensive responses. What is the likely cause?

• Overfitting
• Biased training data
• Gradient vanishing
• Lack of features

7. A logistics company uses AI to predict delivery times. The model struggles with rare events like floods. Which approach helps?

• Feature scaling
• Bagging
• PCA
• Data augmentation

8. A healthcare AI predicts patient readmission. Doctors want to know which features matter most. Which technique is suitable?

• Feature importance analysis
• Gradient descent
• Bagging
• One-hot encoding

9. A stock trading AI makes risky trades during volatile markets. Which reinforcement learning concept can prevent this?

• Gradient vanishing
• Exploration-exploitation balance
• Bagging
• PCA

10. A voice assistant struggles with accents. What technique can improve performance?

• Transfer learning
• Bagging
• PCA
• Logistic regression

11. A retailer uses AI for demand forecasting but finds predictions unstable. Which ensemble method can stabilize results?

• Bagging
• Boosting
• PCA
• Gradient descent

12. A cybersecurity AI detects malware but misses new variants. Which approach improves detection?

• Logistic regression
• PCA
• Anomaly detection
• Bagging

13. A medical AI predicts cancer risk but is criticized for being a “black box.” Which model type offers better interpretability?

• Decision trees
• Deep neural networks
• PCA
• Gradient descent

14. A recommendation system over-prioritizes popular items, ignoring niche products. Which problem is this?

• Popularity bias
• Overfitting
• Data leakage
• Gradient vanishing

15. A financial AI predicts credit scores but unfairly penalizes certain groups. Which principle addresses this?

• Fairness in AI
• Gradient descent
• PCA
• Bagging

16. A weather AI predicts rainfall but ignores seasonal cycles. Which technique can capture periodic patterns?

• Time series modeling with seasonality
• PCA
• Bagging
• Logistic regression

17. A manufacturing AI predicts equipment failure but struggles with imbalanced data (few failures). Which technique helps?

• PCA
• SMOTE (Synthetic Minority Oversampling Technique)
• Bagging
• Gradient descent

18. A translation AI struggles with idioms. Which NLP technique improves understanding?

• Contextual embeddings (e.g., BERT)
• One-hot encoding
• PCA
• Bagging

19. A retail AI predicts customer churn but fails to adapt when customer behavior changes. Which approach helps?

• PCA
• Online learning
• Logistic regression
• Bagging

20. A smart city AI predicts traffic congestion but struggles with sudden road closures. Which technique can handle dynamic changes?

• PCA
• Bagging
• Real-time reinforcement learning
• Logistic regression