AI-900 Practice Test 21 — 15 Questions

Question 1

What is a 'multi-turn conversation' in the context of Azure Bot Service and conversational AI?

Accepted Answer

A conversation where context is maintained across multiple exchanges to enable natural dialogue. In Azure Bot Service and conversational AI, a multi-turn conversation refers to the ability of a bot to maintain context across multiple exchanges, allowing for natural, coherent dialogue. This is achieved through state management (e.g., using Bot Framework's turn context and storage layers) to track user intent and conversation history, enabling follow-up questions and clarifications without losing context.

Answer

A conversation with multiple human agents taking turns responding

Answer

A bot that can respond in multiple languages within one conversation

Answer

A conversation with multiple topics handled simultaneously in parallel

Question 2

A company analyzes customer reviews to identify common themes. They need to automatically extract the most important concepts from each review, such as 'battery life', 'customer service', and 'price'. Which Azure AI Language feature should they use?

Accepted Answer

Key phrase extraction. Key phrase extraction is the correct Azure AI Language feature because it automatically identifies and extracts the most important concepts, such as 'battery life', 'customer service', and 'price', from unstructured text like customer reviews. It returns a list of key phrases that represent the main topics discussed, which directly matches the requirement to extract important concepts.

Answer

Sentiment analysis

Answer

Language detection

Answer

Named entity recognition

Question 3

What is the role of the Azure AI Foundry (AI Studio) playground?

Accepted Answer

An interactive testing environment for experimenting with AI models and prompts without coding. The Azure AI Foundry (AI Studio) playground provides an interactive, no-code environment where developers and data scientists can experiment with generative AI models, test prompts, and adjust parameters like temperature and max tokens before integrating them into applications. This aligns with the need to prototype and validate model behavior without writing code, making it a key tool for rapid iteration in generative AI workloads.

Answer

A gaming environment where AI plays against human developers

Answer

A virtual machine for running AI model training jobs

Answer

A sandbox for testing AI models in isolation from production data

Question 4

What is a training dataset in machine learning?

Accepted Answer

The labeled data used to teach a machine learning model. A training dataset is the labeled data used to teach a machine learning model by allowing it to learn patterns and relationships between features and labels. In Azure Machine Learning, this dataset is fed into an algorithm during the training step, where the model adjusts its internal parameters (e.g., weights in a neural network) to minimize prediction error. Without labeled training data, supervised learning models cannot learn the mapping from inputs to outputs.

Answer

A dataset used to evaluate a trained model's performance on unseen data

Answer

Data that has been cleaned and normalized for analysis

Answer

Real-world data used after model deployment

Question 5

A developer uses Azure OpenAI Service to generate code snippets. They need the model to produce the most likely completion each time, with no randomness or creativity. Which parameter should they set?

Accepted Answer

temperature = 0. Setting temperature = 0 forces the model to always select the token with the highest probability at each step, eliminating randomness and ensuring deterministic, most-likely completions. This is ideal for tasks like code generation where consistency and predictability are required, as it disables the sampling randomness that higher temperature values introduce.

Answer

temperature = 1

Answer

top_p = 0.5

Answer

frequency_penalty = 0.5

Question 6

A digital marketing agency wants to use an AI model that can create original images of products in different styles based on text prompts, such as 'a luxury watch in a futuristic setting.' Which Azure service should they choose?

Accepted Answer

Azure OpenAI Service. Azure OpenAI Service provides access to generative AI models like DALL-E, which can create original images from text prompts. This service is specifically designed for tasks such as generating product images in different styles based on descriptive text, making it the correct choice for the agency's requirement.

Answer

Azure AI Language

Answer

Azure Cognitive Search

Answer

Azure Computer Vision

Question 7

A bank deploys an AI system that uses a deep neural network to approve personal loan applications. A customer whose loan was rejected requests a detailed explanation of why the decision was made. The bank's AI team realizes that the model's internal workings are too complex to provide a simple, understandable reason. According to Microsoft's responsible AI principles, which principle is most directly violated by this situation?

Accepted Answer

Transparency. The bank's inability to provide a clear, understandable explanation for the AI's loan decision directly violates the transparency principle. Microsoft's responsible AI principles require that AI systems be understandable and that their decisions can be explained to users, especially when those decisions have significant impact. A deep neural network's complex, non-linear decision boundaries and lack of inherent interpretability make it a 'black box,' which undermines the required transparency.

Answer

Fairness

Answer

Reliability & Safety

Answer

Privacy & Security

Question 8

A data scientist trains a machine learning model on historical sales data to predict future sales volume. The model achieves 99% accuracy on the training dataset but only 75% accuracy on a separate test dataset. What is the most likely issue with this model?

Accepted Answer

Overfitting. The model's 99% accuracy on the training set versus 75% on the test set indicates it has memorized the training data, including noise and outliers, rather than learning generalizable patterns. This classic symptom of overfitting occurs when the model is too complex relative to the amount or variability of the training data, causing poor performance on unseen data.

Answer

Underfitting

Answer

High bias

Answer

High variance

Question 9

What is 'abstractive summarisation' and how does it differ from 'extractive summarisation'?

Accepted Answer

Extractive selects key sentences verbatim; abstractive generates new sentences capturing the meaning. Extractive summarisation works by selecting and concatenating the most important sentences directly from the source text, while abstractive summarisation uses natural language generation (NLG) models to produce entirely new sentences that paraphrase and condense the core meaning. This distinction is fundamental in Azure AI Language's summarisation capabilities, where extractive returns verbatim excerpts and abstractive generates novel, coherent summaries.

Answer

Extractive writes shorter summaries; abstractive writes longer ones

Answer

Abstractive summarisation is only available for non-English languages

Answer

Extractive summarisation uses generative AI; abstractive uses keyword ranking

Question 10

What does sentiment analysis do?

Accepted Answer

Determines the emotional tone (positive, negative, neutral) expressed in text. Sentiment analysis is a natural language processing (NLP) technique that evaluates text to determine the emotional tone expressed, typically classifying it as positive, negative, or neutral. In Azure Cognitive Services, this is performed by the Text Analytics API, which uses machine learning models to assign sentiment scores and labels based on the overall context of the input text. Option B is correct because it directly describes this core function of detecting emotional polarity.

Answer

Translates text from one language to another

Answer

Extracts key phrases and named entities from text

Answer

Converts spoken words into written text

Question 11

Which Azure service enables building a question-answering chatbot from existing FAQ documents with minimal code?

Accepted Answer

Azure AI Language question answering + Azure Bot Service. Azure AI Language question answering (formerly QnA Maker) is specifically designed to extract question-answer pairs from semi-structured content like FAQ documents, PDFs, and URLs, and expose them via a REST API. Azure Bot Service provides the conversational interface and channel integration (e.g., Teams, Web Chat) with minimal code. Together, they enable a fully functional Q&A chatbot without custom NLP model training.

Answer

Azure Machine Learning with custom NLP models

Answer

Azure AI Speech with custom commands

Answer

Azure Cognitive Search alone

Question 12

A law firm receives hundreds of legal documents daily. They want to use Azure AI Language to automatically assign each document to exactly one predefined category, such as 'Contract', 'Trademark', or 'Litigation'. Which Azure AI Language feature is specifically designed for this task?

Accepted Answer

Custom text classification (single-label). Custom text classification (single-label) is the correct feature because it allows you to train a model to assign each document to exactly one predefined category (e.g., 'Contract', 'Trademark', 'Litigation') based on your own labeled data. This is distinct from prebuilt features like NER or key phrase extraction, which do not perform document-level categorization into custom classes.

Answer

Named Entity Recognition (NER)

Answer

Key phrase extraction

Answer

Conversational Language Understanding (CLU)

Question 13

A hospital wants to automatically anonymize patient medical records by removing all personally identifiable information (PII) such as names, dates, and social security numbers from unstructured text notes. Which Azure AI Language feature should they use?

Accepted Answer

Named entity recognition (NER). Named entity recognition (NER) is the correct Azure AI Language feature because it is specifically designed to identify and categorize entities in unstructured text, including PII categories such as names, dates, and social security numbers. The hospital can use NER's pre-built PII detection model to automatically locate and redact these sensitive elements from patient notes, meeting their anonymization requirement.

Answer

Key phrase extraction

Answer

Sentiment analysis

Answer

Language detection

Question 14

A data science team trains a regression model to predict house prices. They evaluate the model using Mean Absolute Error (MAE). After deployment, they notice that the model occasionally produces large errors (e.g., underpredicting a luxury home by $500,000) while most predictions are within $20,000. The business is more concerned about the impact of these large errors than the average small error. Which additional metric should the team use to better capture the penalty for large errors?

Accepted Answer

Root Mean Squared Error (RMSE). Root Mean Squared Error (RMSE) is the correct additional metric because it squares the residuals before averaging, which heavily penalizes large errors like the $500,000 underprediction. Unlike MAE, which treats all errors equally, RMSE amplifies the impact of outliers, making it a better fit for a business that cares more about catastrophic failures than typical small errors. This aligns with the need to capture the penalty for large deviations in regression model evaluation.

Answer

R-squared

Answer

Mean Absolute Percentage Error (MAPE)

Answer

F1 score

Question 15

What is 'chain of thought' prompting in generative AI?

Accepted Answer

A prompting technique that elicits step-by-step reasoning to improve accuracy on complex tasks. Chain of thought prompting is a technique where the model is asked to produce intermediate reasoning steps before arriving at a final answer, which significantly improves performance on multi-step arithmetic, logic, and common-sense reasoning tasks. Unlike a simple answer request, it forces the model to externalize its reasoning process, reducing errors from shortcut or pattern-matching behaviors. This is a prompting strategy, not a model architecture change, and is particularly effective in large language models like GPT-4 or Azure OpenAI's GPT-3.5 Turbo.

Answer

Connecting multiple AI models in a processing pipeline

Answer

Linking multiple conversation turns to maintain context

Answer

Training a model using sequential text data only