Microsoft Azure AI Fundamentals Artificial Intelligence 2 — Questions and Answers
Question 1: Which type of AI system learns from labeled training data to predict outcomes for new, unseen data?
- Unsupervised learning
- Supervised learning (Correct answer)
- Reinforcement learning
- Generative AI
Correct answer: Supervised learning
Supervised learning uses labeled input-output pairs so the model can learn to predict outputs for new inputs.
Question 2: A company wants to group customers into segments based on purchasing behavior without predefined categories. Which AI approach is most appropriate?
- Supervised classification
- Regression analysis
- Unsupervised clustering (Correct answer)
- Anomaly detection
Correct answer: Unsupervised clustering
Unsupervised clustering identifies natural groupings in data without requiring labeled examples or predefined categories.
Question 3: What is the primary purpose of a validation dataset in machine learning model training?
- To provide the initial training examples
- To evaluate final model performance after deployment
- To tune hyperparameters and detect overfitting during training (Correct answer)
- To generate synthetic training data
Correct answer: To tune hyperparameters and detect overfitting during training
A validation dataset is held out during training to tune hyperparameters and monitor for overfitting before final evaluation.
Question 4: Which term describes a model that performs well on training data but poorly on new, unseen data?
- Underfitting
- Overfitting (Correct answer)
- Regularization
- Normalization
Correct answer: Overfitting
Overfitting occurs when a model memorizes training data patterns rather than learning generalizable rules.
Question 5: In the context of responsible AI, what does the principle of 'reliability and safety' primarily address?
- Ensuring AI systems produce consistent, accurate results and perform safely across conditions (Correct answer)
- Keeping AI model weights private from competitors
- Limiting AI use to critical infrastructure only
- Requiring human approval for every AI decision
Correct answer: Ensuring AI systems produce consistent, accurate results and perform safely across conditions
Reliability and safety means AI systems should behave consistently, handle edge cases safely, and meet the needs they were designed for.
Question 6: What type of bias occurs when training data does not accurately represent the real-world population the model will serve?
- Confirmation bias
- Representation bias (Correct answer)
- Automation bias
- Anchoring bias
Correct answer: Representation bias
Representation bias happens when certain groups or scenarios are underrepresented or overrepresented in training data, skewing predictions.
Question 7: Which Azure service provides a centralized hub for managing the machine learning lifecycle, including experiment tracking and model deployment?
- Azure Cognitive Services
- Azure Machine Learning (Correct answer)
- Azure Databricks
- Azure Synapse Analytics
Correct answer: Azure Machine Learning
Azure Machine Learning is the end-to-end platform for building, training, deploying, and managing ML models with full lifecycle tracking.
Microsoft Azure AI Fundamentals Artificial Intelligence 2
Which type of AI system learns from labeled training data to predict outcomes for new, unseen data?