Microsoft Certified: Azure AI Fundamentals (AI-900) — Questions and Answers

Question 1: What is the purpose of sentiment analysis in Azure AI Language?

Translate text between languages
Summarize long documents
Extract named entities from text
Determine if text expresses positive, negative, or neutral opinion (Correct answer)

Correct answer: Determine if text expresses positive, negative, or neutral opinion

Sentiment analysis evaluates text and assigns a sentiment label (positive, negative, or neutral) along with a confidence score.

Question 2: What is the key difference between Power Virtual Agents and Azure Bot Service?

Power Virtual Agents is a low-code/no-code tool for business users, while Azure Bot Service provides full SDK control for developers (Correct answer)
Azure Bot Service cannot connect to third-party messaging platforms
Power Virtual Agents only integrates with Microsoft Teams, while Azure Bot Service works on all channels
Azure Bot Service requires no coding, while Power Virtual Agents requires extensive programming

Correct answer: Power Virtual Agents is a low-code/no-code tool for business users, while Azure Bot Service provides full SDK control for developers

Power Virtual Agents targets business users with a graphical, low-code interface, while Azure Bot Service (with Bot Framework SDK) gives developers full programmatic control.

Question 3: Which Azure service is primarily used for detecting the language of a text input?

Azure Form Recognizer
Azure Language Detection API in Azure AI Language (Correct answer)
Azure Speech Service
Azure Translator

Correct answer: Azure Language Detection API in Azure AI Language

Azure AI Language includes a language detection feature that identifies the language of submitted text.

Question 4: Which Azure service can analyze text in multiple languages and return the detected language with a confidence score?

Azure AI Search
Azure AI Speech
Azure AI Language – Language Detection (Correct answer)
Azure AI Vision

Correct answer: Azure AI Language – Language Detection

The Language Detection feature within Azure AI Language identifies the language of text input and provides an ISO language code along with a confidence score.

Question 5: Which Azure AI Vision feature generates a human-readable text description of an image?

Image captioning (Correct answer)
OCR
Spatial analysis
Face recognition

Correct answer: Image captioning

Image captioning in Azure AI Vision produces a natural language sentence that describes the content and context of an image.

Question 6: Which NLP concept describes the multiple meanings a single word can have depending on context?

Lemmatization
Lexical ambiguity (Correct answer)
Tokenization
Stemming

Correct answer: Lexical ambiguity

Lexical ambiguity refers to the phenomenon where a word has multiple possible meanings that must be resolved using surrounding context.

Question 7: What is the purpose of Azure Custom Vision?

Provide pre-built models for common vision tasks without training
Allow users to train custom image classification and object detection models (Correct answer)
Detect PII in documents
Transcribe text from audio files

Correct answer: Allow users to train custom image classification and object detection models

Azure Custom Vision lets you train your own image classification and object detection models using your own labeled images without writing ML code.

Question 8: Which Azure service allows you to create a custom speech recognition model tailored to your organization's specific vocabulary?

Azure Text Analytics
Azure Translator
Azure Bot Framework
Azure Custom Speech (part of Speech Services) (Correct answer)

Correct answer: Azure Custom Speech (part of Speech Services)

Azure Custom Speech lets you upload domain-specific audio and transcripts to train a custom speech model that recognizes specialized terminology.

Question 9: Which metric would be most appropriate for evaluating a binary classification model predicting rare fraud events?

Mean Squared Error
Precision-Recall AUC (Correct answer)
Accuracy
R-squared

Correct answer: Precision-Recall AUC

For imbalanced datasets like fraud detection, Precision-Recall AUC is more informative than accuracy, which can be misleadingly high when the positive class is rare.

Question 10: What is 'data poisoning' in the context of AI security?

Intentionally introducing malicious or misleading data into a training dataset to corrupt the model's behavior (Correct answer)
Adding noise to model weights during training
Encrypting training data to prevent unauthorized access
Oversampling minority classes to improve fairness

Correct answer: Intentionally introducing malicious or misleading data into a training dataset to corrupt the model's behavior

Data poisoning is an adversarial attack where malicious data is injected into a training set to manipulate the model's learned behavior in harmful ways.

Question 11: You are creating an AI system that empowers everyone, even those who are visually impaired, hard of hearing, or have other disabilities. Which Microsoft AI guiding principle is exemplified by this situation?

accountability
inclusiveness (Correct answer)
reliability and safety
fairness

Correct answer: inclusiveness

Creating an AI system that empowers everyone, including those with visual impairments, hearing impairments, or other disabilities, directly aligns with the Microsoft AI guiding principle of inclusiveness. This principle focuses on designing AI systems that are accessible and beneficial to the widest possible range of people, regardless of their abilities or background, ensuring equitable access and participation.

Question 12: Which Microsoft tool allows business users and subject-matter experts to create chatbots without writing code?

LUIS Studio
Power Virtual Agents (Correct answer)
Bot Framework SDK
Azure Bot Service

Correct answer: Power Virtual Agents

Power Virtual Agents is a low-code/no-code platform that enables non-developers to build and deploy bots using a guided graphical interface.

Question 13: Which Azure AI service would you integrate with a bot to allow it to understand spoken (voice) input from users?

Azure AI Document Intelligence
Azure AI Vision
Azure AI Speech (speech-to-text) (Correct answer)
Azure AI Translator

Correct answer: Azure AI Speech (speech-to-text)

Azure AI Speech provides speech-to-text capabilities, converting spoken audio into text that the bot can then process as a standard text utterance.

Question 14: Which type of machine learning involves learning from labeled training data to predict outcomes for new inputs?

Semi-supervised learning
Unsupervised learning
Supervised learning (Correct answer)
Reinforcement learning

Correct answer: Supervised learning

Supervised learning uses labeled datasets where the correct outputs are known to train models that can predict outcomes for new, unseen inputs.

Question 15: What forms the core of the personal digital assistant?

Text Analytics
Azure Speech
Computer Vision Service
Bot Framework (Correct answer)

Correct answer: Bot Framework

The Microsoft Bot Framework serves as the core technology stack for building, connecting, and managing intelligent bots, including personal digital assistants. It provides a comprehensive SDK, tools, and services that enable developers to create conversational AI agents capable of understanding natural language, managing dialogue, and integrating with various platforms and services. It is the foundational platform for bot development.

Question 16: What is the function of the 'intent' in a Conversational Language Understanding (CLU) model?

A word-level label for named entities
A confidence threshold for model predictions
The predicted goal or action the user wants to perform (Correct answer)
A translation target for multilingual bots

Correct answer: The predicted goal or action the user wants to perform

In CLU, an intent represents the purpose or goal behind a user's utterance, such as 'BookFlight' or 'CheckWeather'.

Question 17: Which Azure AI Vision feature can generate a dense description of all regions within an image?

Spatial analysis
Smart cropping
Dense captioning (Correct answer)
Object tagging

Correct answer: Dense captioning

Dense captioning generates individual natural language captions for multiple regions across an image, providing a richer description than a single overall caption.

Question 18: Which Azure AI Vision feature can identify the dominant colors and overall mood of an image?

Face detection
Image analysis – color scheme (Correct answer)
Spatial analysis
Object detection

Correct answer: Image analysis – color scheme

The image analysis capability in Azure AI Vision includes color scheme detection, returning dominant foreground/background colors and an accent color.

Question 19: What is a 'tag' in the context of Azure AI Vision image analysis?

A bounding box coordinate for object detection
A metadata label attached to an Azure resource
A keyword describing objects, actions, or concepts visible in an image (Correct answer)
An error code returned when image analysis fails

Correct answer: A keyword describing objects, actions, or concepts visible in an image

Azure AI Vision returns tags — descriptive keywords with confidence scores that identify recognizable objects, scenes, and actions present in an image.

Question 20: Find the Azure cloud service that includes serverless features.

Azure App Service
Azure Functions
Knowledge, Translation, Audio, Visual and Speech
All of the above (Correct answer)
Azure Kubernetes Service (AKS)

Correct answer: All of the above

Azure offers multiple services with serverless capabilities. Azure Functions is its primary serverless compute service, but Azure App Service can host serverless web apps, and Azure Kubernetes Service (AKS) can integrate with serverless components through KEDA or by running serverless-like workloads. Therefore, all listed options can include serverless features.

Question 21: There are _______ proposition symbols in total.

No proposition symbols
3 proposition symbols
2 proposition symbols (Correct answer)
1 proposition symbols

Correct answer: 2 proposition symbols

In propositional logic, a proposition symbol (or atomic proposition) represents a basic statement that can be either true or false. Therefore, any given proposition symbol can take on one of two possible truth values: True or False. The number '2' refers to these two fundamental truth values that define the state of any proposition in classical logic.

Question 22: According to Microsoft's Responsible AI framework, who bears responsibility for ensuring AI systems are used ethically?

Only the government agencies that regulate AI
End users who interact with the AI
Both the organizations that build and deploy AI systems (Correct answer)
Only the AI model developers who build the system

Correct answer: Both the organizations that build and deploy AI systems

The Accountability principle places responsibility on all organizations involved — both those that build AI systems and those that deploy them — to ensure ethical use.

Question 23: In Azure conversational AI, what does LUIS stand for?

Linguistic Understanding Integration System
Large Unified Intent System
Language Understanding Intelligent Service (Correct answer)
Language Unified Intelligence Suite

Correct answer: Language Understanding Intelligent Service

LUIS stands for Language Understanding Intelligent Service, a cloud-based API service that applies machine learning to natural language to predict meaning and extract information.

Question 24: Which Responsible AI principle focuses on ensuring AI benefits and is accessible to all people, regardless of ability?

Transparency
Reliability & Safety
Inclusiveness (Correct answer)
Privacy & Security

Correct answer: Inclusiveness

The Inclusiveness principle requires AI systems to be designed to empower everyone, including people with disabilities, and avoid excluding groups of people.

Question 25: A company wants to detect whether a customer review is positive, negative, or neutral. Which AI technique is most appropriate?

Sentiment analysis (Correct answer)
Image classification
Object detection
Speech synthesis

Correct answer: Sentiment analysis

Sentiment analysis is an NLP technique that determines the emotional tone of text, categorizing it as positive, negative, or neutral.

Question 26: Which Microsoft Responsible AI principle requires that AI systems work correctly and safely even in unexpected situations?

Reliability and Safety (Correct answer)
Accountability
Privacy and Security
Inclusiveness

Correct answer: Reliability and Safety

The Reliability and Safety principle mandates that AI systems perform as intended, handle failure gracefully, and avoid causing harm in unanticipated scenarios.

Question 27: Using _____________________, you may extract the information from food product labels.

image segmentation
optical character recognition (Correct answer)
image classification
natural language processing

Correct answer: optical character recognition

Optical Character Recognition (OCR) is the technology specifically used to extract text from images or scanned documents, converting it into a machine-readable format. Food product labels contain printed text, such as ingredients, nutritional information, and brand names. Therefore, OCR is the appropriate method to digitize and extract this textual information for further processing or analysis.

Question 28: Which Azure AI service can analyze videos to extract insights like speakers, topics, and transcript?

Azure AI Video Indexer (Correct answer)
Azure AI Vision
Azure AI Document Intelligence
Azure AI Speech

Correct answer: Azure AI Video Indexer

Azure AI Video Indexer automatically extracts insights from videos including transcripts, speaker identification, sentiment, topics, and key frames.

Question 29: In Azure AI Vision, what is 'object detection' different from 'image classification'?

Object detection returns text; classification returns coordinates
Object detection only works on videos, classification works on images
Object detection requires custom training; classification uses pre-built models only
Object detection identifies and locates multiple objects with bounding boxes; classification assigns a single label to the whole image (Correct answer)

Correct answer: Object detection identifies and locates multiple objects with bounding boxes; classification assigns a single label to the whole image

Object detection finds and locates multiple objects in an image with bounding boxes, while image classification assigns a single category label to the entire image.

Question 30: What is the purpose of a 'model card' in Responsible AI documentation?

A card game used to teach AI concepts to students
A credit card used to pay for Azure AI services
A structured document that describes a model's intended use, performance metrics, limitations, and ethical considerations (Correct answer)
A business card format for AI researchers to share contact information

Correct answer: A structured document that describes a model's intended use, performance metrics, limitations, and ethical considerations

A model card is a transparency artifact that summarizes what a model does, how it performs across different groups, its limitations, and the contexts in which it should or should not be used.

Question 31: Which iteration-based workflow does Azure Custom Vision use to improve model accuracy?

Upload all images at once and deploy without testing
Generate synthetic data automatically without user labels
Use pre-trained ImageNet weights without any customization
Active learning loop: label → train → evaluate → retrain with corrected images (Correct answer)

Correct answer: Active learning loop: label → train → evaluate → retrain with corrected images

Azure Custom Vision follows an iterative cycle where you label images, train, evaluate performance, correct errors, and retrain to progressively improve accuracy.

Question 32: All of the information listed here, with the exception of:

Regions
Areas (Correct answer)
Words
Areas
Lines

Correct answer: Areas

This question, likely referring to document processing services like Azure Form Recognizer, asks what information is *not* typically listed or extracted. Services like Form Recognizer are designed to extract structured information such as individual words, lines of text, and logical regions (like paragraphs or fields). 'Areas' is a very general term and doesn't represent a specific, actionable type of information extracted in the same granular way as words, lines, or defined regions.

Question 33: You must create a smartphone application so that employees may scan and store their trip expenses. Which kind of computer vision ought to be employed?

object detection
optical character recognition (OCR) (Correct answer)
semantic segmentation
image classification

Correct answer: optical character recognition (OCR)

Scanning and storing trip expenses from physical documents like receipts involves converting images of text into machine-readable text. Optical Character Recognition (OCR) is the specific computer vision technology designed for this purpose. OCR allows the application to extract textual information from the scanned images, making it searchable and usable for expense tracking.

Question 34: What is the recommended approach for handling user input that the bot does not recognize or understand?

Immediately end the conversation and log an error
Implement fallback handling to gracefully respond and guide the user back on track (Correct answer)
Redirect the user to an external FAQ website without responding
Stop responding until the user rephrases the message correctly

Correct answer: Implement fallback handling to gracefully respond and guide the user back on track

Fallback handling ensures the bot responds gracefully to unrecognized input — for example, by apologizing, offering menu options, or escalating to a human agent.

Question 35: Which Microsoft Responsible AI principle states that AI systems should be understandable and their decisions explainable?

Accountability
Transparency (Correct answer)
Fairness
Reliability

Correct answer: Transparency

The Transparency principle requires that AI systems and their decision-making processes be understandable and explainable to users and stakeholders.

Question 36: Azure Personalizer uses which type of machine learning to deliver personalized content recommendations?

Reinforcement learning (Correct answer)
Unsupervised learning
Supervised learning
Transfer learning

Correct answer: Reinforcement learning

Azure Personalizer uses reinforcement learning, where it learns which actions (content selections) yield the best rewards (user engagement) over time.

Question 37: What does the term 'tokenization' mean in the context of NLP?

Encrypting text data for security
Breaking text into individual words or subword units (Correct answer)
Removing stop words from a sentence
Translating text into a numeric format

Correct answer: Breaking text into individual words or subword units

Tokenization is the process of splitting text into smaller units called tokens, which are typically words or subword pieces.

Question 38: Which capability allows Azure AI Language to identify the main topics discussed in a document?

Entity linking
Sentiment analysis
Language detection
Key phrase extraction (Correct answer)

Correct answer: Key phrase extraction

Key phrase extraction identifies the most important phrases that capture the main topics within a document.

Question 39: In the Bot Framework, what is a 'dialog'?

A text response generated and returned by the bot to the user
A modular, reusable component that manages a specific part of the conversation flow (Correct answer)
A LUIS feature for grouping related utterances together
A channel connector used to link the bot to communication platforms

Correct answer: A modular, reusable component that manages a specific part of the conversation flow

A dialog is a self-contained component in Bot Framework that encapsulates a specific conversational task (e.g., collecting a booking date), enabling reuse and composability.

Question 40: What does 'stemming' accomplish in NLP preprocessing?

Converts words to their root or base form by removing suffixes (Correct answer)
Translates words to another language
Splits text into individual characters
Identifies named entities in a sentence

Correct answer: Converts words to their root or base form by removing suffixes

Stemming reduces words to their base form (e.g., 'running' → 'run') by stripping affixes, helping normalize text for analysis.

Microsoft Certified: Azure AI Fundamentals (AI-900)

This exam validates foundational knowledge of machine learning and artificial intelligence (AI) concepts and related Microsoft Azure AI services.

Questions

60 min

Time Limit

70%

To Pass

Checking for saved progress...

Exam Rules

You can skip questions and return to them later
Flag questions for review before submitting
No feedback shown until you submit the entire exam
Unanswered questions count as wrong — answer everything
10 pretest questions are mixed in and don't affect your score
Timer auto-submits when time runs out
Your progress is auto-saved every 30 seconds

These practice questions were written with AI tools. They are not real exam questions. Exam facts are checked against the official sources listed below. Not reviewed by a subject-matter expert. Editorial policy

Sources & references

Exam facts on this page (fees, format, dates, passing scores) are checked against these official sources; the date is our last check. Editorial policy

learn.microsoft.com — Renewal, score validity, test delivery, retake fee, retake policy, exam format, time limit, passing score, exam information · verified Nov 2026