Pass the Microsoft Microsoft Certified: Azure AI Fundamentals AI-900 Questions and answers with CertsForce

According to the Microsoft Azure OpenAI Service documentation and AI-900 official study materials, the DALL-E model is specifically designed to generate and edit images from natural language prompts. When a user provides a descriptive text input such as “a futuristic city skyline at sunset”, DALL-E interprets the textual prompt and produces an image that visually represents the content described. This functionality is known as text-to-image generation and is one of the creative AI capabilities supported by Azure OpenAI.

DALL-E belongs to the family of generative models that can create new visual content, expand existing images, or apply transformations to images based on textual instructions. Within Azure OpenAI, the DALL-E API enables developers to integrate image creation directly into applications—useful for design assistance, marketing content generation, or visualization tools. The model learns from vast datasets of text–image pairs and is optimized to ensure alignment, diversity, and accuracy in the produced visuals.

By contrast, the other options serve different purposes:

A. GPT-4 is a large language model for text-based generation, reasoning, and conversation, not for creating images.

C. GPT-3 is an earlier text generation model, primarily used for language tasks like summarization, classification, and question answering.

D. Whisper is an automatic speech recognition (ASR) model used to convert spoken language into written text; it has no image-generation capability.

Therefore, when the requirement is to generate images based on user prompts, the only Azure OpenAI model that fulfills this purpose is DALL-E. This aligns directly with the AI-900 learning objective covering Azure OpenAI generative capabilities for text, code, and image creation.

According to the Microsoft Azure AI Fundamentals (AI-900) official study guide and the Microsoft Learn module “Identify features of common machine learning types”, the three main machine learning model types differ by their purpose and the kind of data they use — whether supervised (using labeled data) or unsupervised (using unlabeled data).

Regression → A supervised machine learning model used to predict numeric values.Regression is a type of supervised learning that predicts continuous numerical outcomes. It learns the relationship between input features (independent variables) and a continuous target variable (dependent variable). Examples include predicting house prices, sales revenue, or temperature. The AI-900 curriculum emphasizes regression for “predicting numeric values based on known data,” using algorithms such as linear regression or decision tree regression.

Classification → A supervised machine learning model used to predict categories.Classification is also a supervised learning technique, but it predicts discrete outcomes (classes) instead of continuous values. It assigns input data to one or more categories based on learned patterns. Typical examples include spam detection (spam vs. not spam), sentiment analysis (positive, neutral, negative), or predicting loan approval (approved/denied). The AI-900 study materials describe classification as “predicting a category or label for new observations.”

Clustering → An unsupervised machine learning model used to group similar entities based on features.Clustering is an unsupervised learning approach, meaning it works on unlabeled data. It automatically identifies patterns and groups similar data points into clusters based on shared characteristics. Examples include customer segmentation (grouping customers by behavior) and grouping similar documents. The AI-900 learning module explains clustering as “discovering natural groupings in data without predefined labels.”

Thus, per Microsoft’s official AI-900 learning objectives:

Regression → Predicts numeric/continuous values.

Classification → Predicts categories/labels.

Clustering → Groups similar entities (unsupervised).

✅ Final Verified Configuration:

Regression → Predict numeric values

Classification → Predict categories

Clustering → Group similar entities based on features

The correct answer is B. Optical Character Recognition (OCR).

Optical Character Recognition (OCR) is a feature of Azure AI Vision that converts printed or handwritten text within images into machine-readable text. In this scenario, the goal is to read runner numbers on shirts from race photos. OCR can identify and extract these numbers, allowing them to be associated with specific participants.

Option analysis:

A. Image classification: Categorizes entire images (e.g., “runner,” “crowd”), not text.

B. Optical Character Recognition (OCR) — ✅ Correct. Extracts alphanumeric text from images.

C. Object detection: Identifies and locates objects (e.g., shoes, cars) but doesn’t read text.

D. Facial recognition: Identifies individuals by matching facial features to known identities, not by reading numbers.

Therefore, to read and extract runner numbers from photos, the correct computer vision technique is Optical Character Recognition (OCR).

According to the AI-900 official study guide and Microsoft Learn module “Describe features of conversational AI workloads on Azure”, conversational AI refers to artificial intelligence systems that interact with users through natural language via text or speech. These systems include chatbots, virtual assistants, and interactive voice response (IVR) systems that simulate human conversation.

B. Chatbot that provides users with the ability to find answers on a website by themselvesThis is a classic example of conversational AI. Chatbots use natural language understanding (LUIS) and Azure Bot Service to interpret user input, identify intent, and provide relevant responses automatically. They help users self-serve information without human support, such as retrieving account details or answering FAQs.

C. Telephone voice menus to reduce the load on human resourcesAutomated telephone systems or IVRs use conversational AI to interpret spoken commands and route calls intelligently. This is often implemented using Azure Cognitive Services Speech (for speech-to-text and text-to-speech) combined with Azure Bot Service for managing dialogue flow.

Within Azure AI Document Intelligence (formerly Form Recognizer), the Contract prebuilt model is designed to extract key information from legal and business contracts, including parties, jurisdictions, dates, and terms. According to Microsoft Learn, this prebuilt model identifies structured entities such as contracting parties, effective dates, governing jurisdictions, and termination clauses.

Layout (B) extracts text, tables, and structure but does not identify semantic information such as parties or jurisdictions.

General document (C) extracts key-value pairs and entities but lacks domain-specific contract analysis.

Read (D) performs OCR (optical character recognition) to extract raw text but not contextual metadata.

Thus, when the requirement is to extract parties and jurisdictions from a legal document, the Contract model is the correct Azure AI Document Intelligence choice.

According to the Microsoft Azure AI Fundamentals (AI-900) curriculum and Microsoft Learn module “Explore natural language processing”, NLP enables computers to understand, interpret, and analyze human language. One of its key capabilities is sentiment analysis, which detects emotional tone (positive, negative, or neutral) in text.

In this scenario, the chatbot must detect when a customer is upset based on what they type. This directly maps to sentiment analysis, a core NLP function. The Text Analytics service within Azure Cognitive Services provides prebuilt sentiment analysis models that return a sentiment score and classification (e.g., positive, neutral, negative). As per Microsoft Learn, “Natural language processing allows systems to understand sentiment and intent within text and speech to derive meaningful insights.”

Explanation of other options:

A. Anomaly detection identifies unusual patterns in data (e.g., fraud detection), not emotions in text.

B. Semantic segmentation is a computer vision technique used to label pixels in an image.

C. Regression predicts continuous numeric values and is not related to understanding text or emotions.

Therefore, to enable the chatbot to detect when a user is upset based on text input, the correct AI workload is Natural Language Processing (NLP), specifically through Azure Text Analytics sentiment analysis.

In Microsoft’s Responsible AI principles, Fairness ensures that AI systems treat all individuals and groups equitably and make unbiased decisions. The AI-900 study guide explicitly states that fairness is violated when an AI model produces outcomes that systematically favor one group over another — such as preferring a particular gender, race, or age group.

In this scenario, a loan approval system shows gender bias, meaning it approves or rejects applications differently based on gender. This directly contradicts the fairness principle, as the AI system must make decisions solely based on relevant financial attributes (e.g., credit score, income) rather than personal characteristics.

Other principles explained in the AI-900 course include:

Accountability: Ensures human oversight and responsibility.

Transparency: Ensures users understand how decisions are made.

Reliability and Safety: Ensures consistent and accurate operation.

Since gender bias undermines equitable treatment, the principle violated is Fairness.

According to the AI-900 Microsoft Learn modules on Conversational AI, a custom question answering solution built using Azure AI Language (formerly QnA Maker) enables a chatbot to respond to user questions based on a predefined knowledge base. When integrated with a bot, the solution can automatically respond to multiple user queries in real time without additional programming.

This capability is known as scalability and concurrency, which allows chatbots to manage simultaneous conversations with multiple users. This feature is built into the Azure Bot Service, meaning you don’t need to add extra “skills” or custom logic for concurrent interactions.

Other options require additional integration or logic:

Register customer complaints or purchases would require connecting the bot to a CRM or sales system.

Provide RMA numbers requires business process logic or database access.

Therefore, the out-of-the-box functionality of a custom question answering bot is the ability to answer questions from multiple users at once, which is supported natively by Azure Bot Service and the QnA knowledge base.

The scenario requires two checks on each photo: (1) there is at least one face, and (2) at least one detected face is wearing sunglasses. The Azure AI Face service – Detect operation is purpose-built for this combination. It detects faces and returns per-face attributes, including glasses type, so you can enforce both rules in a single pass. From the official guidance, the Detect API “detects human faces in an image and returns the rectangle coordinates of their locations” and exposes face attributes such as glasses. A concise attribute extract states: “Glasses: NoGlasses, ReadingGlasses, Sunglasses, Swimming Goggles.” With this, you can count faces (requirement 1) and then verify that at least one face’s glasses attribute equals sunglasses (requirement 2).

By contrast, other options don’t align as precisely:

A. Verify (Face service) compares whether two detected faces belong to the same person. It does not provide content attributes like sunglasses; it requires face inputs for identity/one-to-one scenarios, which doesn’t meet your content-filter goal.

C. Describe Image (Computer Vision) returns a natural-language caption of the whole image. While a caption might mention “a person wearing sunglasses,” it’s not guaranteed, is not face-scoped, and offers less deterministic filtering than a structured attribute on a detected face.

D. Analyze Image (Computer Vision) can return tags such as “person” or sometimes “sunglasses,” but those tags are image-level and not bound to specific faces. You need to ensure that a detected face (not just any region) is wearing sunglasses. Face-scoped attributes from Face Detect are more reliable for this logic.

Therefore, the most accurate and exam-aligned choice is B. the Detect operation in the Face service, because it allows you to programmatically confirm face presence and per-face sunglasses in a precise, rule-driven workflow.