Microsoft Azure AI Fundamentals Artificial Intelligence 5 — Questions and Answers
Question 1: Which term describes the phenomenon where a generative AI model produces confident but factually incorrect or fabricated information?
- Overfitting
- Hallucination (Correct answer)
- Underfitting
- Prompt injection
Correct answer: Hallucination
Hallucination occurs when a generative AI model generates plausible-sounding but factually incorrect or entirely invented content.
Question 2: A model trained on data from 2020 is used in 2026 without retraining. Which issue is most likely to affect its accuracy?
- Model drift due to data distribution changes over time (Correct answer)
- Overfitting to the 2026 test set
- Feature leakage from the validation set
- Batch normalization errors
Correct answer: Model drift due to data distribution changes over time
Model drift (or data drift) occurs when real-world data patterns change over time, causing a model trained on older data to become less accurate.
Question 3: In the context of Azure AI services, what is the purpose of an API key?
- To encrypt data stored in Azure Blob Storage
- To authenticate requests to an Azure AI service endpoint (Correct answer)
- To define the model architecture for training
- To configure virtual network routing for AI workloads
Correct answer: To authenticate requests to an Azure AI service endpoint
An API key authenticates API calls to Azure AI services, ensuring only authorized applications can access the resource.
Question 4: Which responsible AI principle requires that people and organizations be answerable for the decisions and actions of AI systems?
- Transparency
- Fairness
- Inclusiveness
- Accountability (Correct answer)
Correct answer: Accountability
Accountability means humans and organizations must take responsibility for AI system outcomes and have processes to address harms.
Question 5: What is the difference between artificial intelligence and machine learning?
- They are identical terms used interchangeably in industry
- AI is the broader field of creating intelligent systems; ML is a subset where systems learn from data (Correct answer)
- Machine learning is broader than AI and includes rule-based systems
- AI requires labeled data while ML does not
Correct answer: AI is the broader field of creating intelligent systems; ML is a subset where systems learn from data
AI encompasses all techniques for creating intelligent behavior in machines, while ML is a specific AI approach where models learn patterns from data.
Question 6: Which Azure service is specifically designed to detect, assess, and mitigate harmful content in AI applications?
- Azure Monitor
- Azure Content Safety (Correct answer)
- Azure Policy
- Azure Sentinel
Correct answer: Azure Content Safety
Azure Content Safety detects harmful or inappropriate content such as hate speech, violence, and sexual content in text and images.
Question 7: A development team wants to monitor their deployed Azure Machine Learning model for changes in prediction accuracy over time. Which concept describes what they are monitoring?
- Hyperparameter tuning
- Model drift (Correct answer)
- Feature engineering
- Cross-validation
Correct answer: Model drift
Model drift monitoring tracks whether a deployed model's performance degrades over time as real-world data patterns evolve.
Microsoft Azure AI Fundamentals Artificial Intelligence 5
Which term describes the phenomenon where a generative AI model produces confident but factually incorrect or fabricated information?