Microsoft Azure AI Fundamentals Microsoft Azure AI Fundamentals 5 — Questions and Answers
Question 1: Which Azure AI capability would you use to enable a call center application to automatically convert customer speech into text for analysis?
- Text-to-Speech
- Speech-to-Text (Speech Recognition) (Correct answer)
- Speaker Recognition
- Language Detection
Correct answer: Speech-to-Text (Speech Recognition)
Speech-to-Text transcribes spoken audio into written text in real time, making it ideal for call center transcription and analysis.
Question 2: What is 'transfer learning' in machine learning?
- Moving a trained model from one Azure region to another
- Applying knowledge learned from one task to improve performance on a different but related task (Correct answer)
- Transferring data between on-premises storage and Azure
- Copying model weights to a new neural network with the same architecture
Correct answer: Applying knowledge learned from one task to improve performance on a different but related task
Transfer learning reuses a pre-trained model's learned representations as a starting point for a new task, reducing training time and data requirements.
Question 3: Which responsible AI principle focuses on ensuring that AI system decisions can be understood and explained to users?
- Fairness
- Inclusiveness
- Transparency (Correct answer)
- Accountability
Correct answer: Transparency
Transparency means AI systems should be understandable — users and stakeholders should be able to see how and why decisions were made.
Question 4: What is the purpose of the 'Content Moderator' service in Azure Cognitive Services?
- To generate new content using generative AI
- To detect potentially offensive, risky, or unwanted content in text and images (Correct answer)
- To summarize long documents into key points
- To classify content into predefined business categories
Correct answer: To detect potentially offensive, risky, or unwanted content in text and images
Azure Content Moderator scans text, images, and video for explicit, offensive, or otherwise policy-violating content to support safe online platforms.
Question 5: In Azure Machine Learning, what is the difference between 'inference' and 'training'?
- Training uses GPUs; inference uses CPUs only
- Training builds the model by learning from data; inference uses the trained model to make predictions (Correct answer)
- Training is done in the cloud; inference is done on-premises
- Training requires labeled data; inference requires unlabeled data only
Correct answer: Training builds the model by learning from data; inference uses the trained model to make predictions
Training is the process of fitting a model to data to learn patterns, while inference is using that trained model to generate predictions on new input data.
Question 6: Which Azure OpenAI model capability allows it to understand and respond to both text and images in the same prompt?
- Embedding generation
- Multi-modal (vision) capability (Correct answer)
- Fine-tuning
- Function calling
Correct answer: Multi-modal (vision) capability
Multi-modal models like GPT-4o can accept both text and image inputs, enabling use cases like image description, visual Q&A, and document analysis.
Question 7: What does 'AutoML' in Azure Machine Learning automatically handle for the user?
- Deploying models to production endpoints
- Iterating through algorithms and hyperparameters to find the best model for a dataset (Correct answer)
- Labeling training data using human reviewers
- Monitoring models for data drift after deployment
Correct answer: Iterating through algorithms and hyperparameters to find the best model for a dataset
AutoML automates the process of algorithm selection, feature engineering, and hyperparameter tuning to find the best-performing model with minimal manual effort.
Microsoft Azure AI Fundamentals Microsoft Azure AI Fundamentals 5
Which Azure AI capability would you use to enable a call center application to automatically convert customer speech into text for analysis?