View Full Microsoft AI-102 Exam Dumps and Practice Test Dumps.

 

Question 1

Which Microsoft service is most appropriate for analyzing the sentiment of customer reviews?

1. Azure AI Search
2. Azure Speech
3. Azure AI Language
4. Azure AI Vision

Correct Answer: 3

Explanation

Azure AI Language provides natural language processing capabilities that can analyze text for sentiment. A customer review can be submitted to the service to determine whether its overall tone is positive, negative, or neutral, and confidence scores can also be returned. Azure AI Search focuses on search and knowledge mining, Azure Speech handles speech-related workloads, and Azure AI Vision processes images and video. Therefore, when the primary requirement is determining the sentiment expressed in written customer feedback, Azure AI Language is the appropriate Microsoft AI service for the solution.

Question 2

A company needs to extract text from scanned invoices and identify fields such as invoice number and total amount. Which service should be used?

1. Azure AI Document Intelligence
2. Azure AI Search
3. Azure Speech
4. Azure AI Vision

Correct Answer: 1

Explanation

Azure AI Document Intelligence is designed to extract structured information from documents. It provides prebuilt models that can process common document types and return fields such as invoice numbers, dates, vendor information, and totals. It can also support custom models when an organization’s documents contain specialized layouts or fields. Azure AI Search is primarily used for indexing and querying content, while Azure Speech focuses on audio and speech processing. Azure AI Vision can perform optical character recognition, but Document Intelligence is more suitable when the requirement includes extracting structured fields from business documents.

Question 3

What is a primary purpose of a Microsoft Foundry project?

1. Store only image files
2. Organize AI development resources
3. Replace Azure subscriptions
4. Provide physical GPU hardware

Correct Answer: 2

Explanation

A Microsoft Foundry project provides an organized environment for developing and managing AI solutions. It can bring together models, connections, evaluations, and other resources needed for an AI application. Projects help teams manage work associated with particular solutions while operating within the broader Microsoft Foundry environment. A project does not replace an Azure subscription, provide physical hardware, or function solely as image storage. When designing an AI solution, understanding how projects organize development resources helps engineers establish an appropriate structure for model deployment, application integration, evaluation, and operational management.

Question 4

An application must convert spoken customer questions into text before sending them to an AI model. Which capability is required?

1. Text translation
2. Image analysis
3. Speech-to-text
4. Text summarization

Correct Answer: 3

Explanation

Speech-to-text converts spoken audio into written text. This capability is useful when an application receives voice input but downstream processing requires textual content. For example, a customer-service application can capture a spoken question, use Azure Speech to convert the audio into text, and then pass the resulting text to a language model or other natural language service. Text translation changes content from one language to another, image analysis processes visual information, and summarization reduces existing text. Therefore, speech-to-text is the appropriate capability for converting spoken questions into text.

Question 5

Which principle helps ensure an AI solution is designed to reduce unintended harm and support appropriate use?

1. Responsible AI
2. Data duplication
3. Resource tagging
4. Network peering

Correct Answer: 1

Explanation

Responsible AI provides principles and practices for developing and operating AI systems in an appropriate and trustworthy manner. When planning an Azure AI solution, engineers should consider issues such as fairness, reliability, safety, privacy, security, inclusiveness, transparency, and accountability. These considerations can influence data selection, model evaluation, access controls, monitoring, and application behavior. Responsible AI is not simply a final testing step; it should be incorporated throughout the solution lifecycle. Applying these principles helps organizations identify potential risks and establish governance practices before and after an AI system is deployed.

Question 6

A generative AI application must answer questions using information stored in company documents. What pattern should the engineer consider?

1. Batch processing
2. RAG
3. Image classification
4. Speech synthesis

Correct Answer: 2

Explanation

Retrieval-augmented generation, or RAG, combines information retrieval with generative AI. Instead of relying only on the model’s previously learned information, the application retrieves relevant content from an organization’s data and provides that content as grounding information for the model. This approach can help generate responses based on current or organization-specific information. A RAG solution commonly involves preparing documents, indexing content, retrieving relevant information, and supplying the retrieved context to a model. Batch processing, image classification, and speech synthesis do not directly address grounding generated responses in enterprise documents.

Question 7

What is the main purpose of prompt engineering in a generative AI solution?

1. Increase storage capacity
2. Configure network routing
3. Improve model responses
4. Create database indexes

Correct Answer: 3

Explanation

Prompt engineering involves designing and refining instructions and input context provided to a generative AI model. A well-designed prompt can clarify the task, establish the desired format, provide relevant context, and place useful constraints on the response. This can improve the consistency and usefulness of model outputs. Prompt engineering does not increase storage capacity, configure network routing, or create database indexes. In an AI-102 solution, prompt engineering is particularly important when an application needs predictable responses, specific output structures, or behavior aligned with business requirements.

Question 8

An organization wants an AI system to autonomously perform tasks by using tools and responding to user requests. What should the solution use?

1. An agent
2. A static image
3. A search index only
4. A speech synthesizer

Correct Answer: 1

Explanation

An AI agent is designed to perform tasks based on instructions, context, available tools, and user interactions. Agents can be configured to reason about a request and take actions through connected capabilities rather than simply returning a fixed response. Microsoft Foundry Agent Service can be used to create and manage agents for supported scenarios. A search index can provide information but does not by itself represent an autonomous task-oriented agent. Similarly, image and speech services provide specific capabilities rather than the orchestration behavior associated with an agentic solution.

Question 9

Which capability is provided by Azure AI Vision for an image containing several objects?

1. Object detection
2. Database replication
3. Speech translation
4. Text generation

Correct Answer: 1

Explanation

Object detection identifies objects within an image and can provide information about their locations using bounding regions. This differs from simple image classification, which generally assigns labels to the image without necessarily identifying where each object appears. Azure AI Vision provides computer vision capabilities that can be used to analyze images for objects and other visual characteristics. Database replication is a storage capability, speech translation handles spoken language, and text generation belongs to generative AI scenarios. Therefore, object detection is the appropriate capability when an application needs to identify and locate multiple objects in an image.

Question 10

A developer needs to convert generated text into natural-sounding spoken audio. Which capability should be selected?

1. OCR
2. Text-to-speech
3. Sentiment analysis
4. Entity recognition

Correct Answer: 2

Explanation

Text-to-speech converts written text into synthesized spoken audio. Azure Speech provides text-to-speech capabilities that can be integrated into applications such as virtual assistants, accessibility tools, customer-service systems, and interactive applications. OCR extracts text from images or documents, sentiment analysis determines the emotional tone of text, and entity recognition identifies items such as people, organizations, or locations. When an application already has text and needs to present that information as spoken output, text-to-speech is the appropriate capability. Speech synthesis can also be customized through supported options for voice and output behavior.

Question 11

Which Azure AI Search component is responsible for retrieving and processing data from a configured data source?

1. Model deployment
2. Agent
3. Indexer
4. Prompt template

Correct Answer: 3

Explanation

An Azure AI Search indexer automates the process of retrieving data from a supported data source and populating or updating a search index. This is useful when organizations need to regularly ingest content from sources such as storage accounts or supported databases. The index stores searchable representations of the content, while the indexer performs the data ingestion process. Model deployments, agents, and prompt templates serve different purposes in AI solutions. Understanding the relationship between data sources, indexers, indexes, and skillsets is important when implementing knowledge-mining solutions with Azure AI Search.

Question 12

A team wants search results to consider the meaning of a user’s query rather than relying only on exact keyword matches. Which capability is most relevant?

1. Semantic search
2. File projection
3. OCR
4. Speech recognition

Correct Answer: 1

Explanation

Semantic search improves search relevance by using language understanding to interpret the meaning and context of queries and documents. This can produce more useful results when the words in a query do not exactly match the wording of relevant content. Azure AI Search supports semantic search capabilities that can enhance traditional keyword-based retrieval. File projections are associated with Knowledge Store outputs, while OCR extracts text from images and speech recognition converts spoken language into text. Therefore, semantic search is the appropriate capability when the goal is to improve result relevance based on the meaning of content.

Question 13

A solution must identify names of people and organizations within a collection of documents. Which NLP capability should be used?

1. Language detection
2. Entity recognition
3. Speech synthesis
4. Image segmentation

Correct Answer: 2

Explanation

Entity recognition identifies important entities in text, such as people, organizations, locations, dates, and other supported categories. This capability can help applications extract structured information from unstructured documents. For example, a document-processing application could identify a company name and the names of individuals mentioned in a contract. Language detection determines which language a text is written in, speech synthesis produces audio, and image segmentation is a computer vision technique. Therefore, entity recognition is the appropriate natural language processing capability when the requirement is to identify names and organizations within textual content.

Question 14

What should an engineer evaluate when selecting a generative AI model for a production application?

1. Only the model name
2. Only the subscription name
3. Task requirements and model capabilities
4. Only the resource region

Correct Answer: 3

Explanation

Model selection should be based on the requirements of the application and the capabilities of the available models. Important considerations can include the type of task, supported input and output modalities, quality requirements, context needs, latency, throughput, cost, and deployment availability. Selecting a model solely because of its name, subscription, or geographic region does not provide enough information to determine whether it is appropriate. A production solution should match model capabilities with the intended workload and evaluate performance against defined requirements. This approach supports better technical and operational decisions when deploying generative AI applications.

Question 15

An application needs to detect whether text contains personally identifiable information. Which capability is appropriate?

1. PII detection
2. Object detection
3. Image captioning
4. Speech synthesis

Correct Answer: 1

Explanation

PII detection identifies potentially sensitive personally identifiable information within text. Examples can include names, addresses, phone numbers, identification numbers, and other supported categories of personal information. This capability can be useful when applications process customer communications, documents, or other text that may contain sensitive data. Object detection is used for identifying objects in images, image captioning generates descriptions of visual content, and speech synthesis converts text into spoken audio. Using PII detection can help organizations identify sensitive content so that appropriate handling, redaction, or protection processes can be applied.

Question 16

Which feature allows an application to provide an image-generation request to a generative AI model?

1. DALL-E
2. Azure AI Search
3. Azure Document Intelligence
4. Azure Speech

Correct Answer: 1

Explanation

DALL-E is a generative model designed for creating images from textual prompts. It can be used in applications where users provide descriptions and the system generates corresponding visual content. Azure AI Search is intended for search and retrieval workloads, Document Intelligence extracts information from documents, and Azure Speech provides speech-related capabilities. When an AI application specifically requires generated images based on natural-language descriptions, an image-generation model such as DALL-E is relevant. Engineers should still consider model availability, deployment requirements, responsible AI considerations, and application-specific requirements when implementing this capability.

Question 17

A document-processing pipeline needs to extract text from scanned images before further analysis. What capability should be included?

1. OCR
2. Text-to-speech
3. Sentiment scoring
4. Model fine-tuning

Correct Answer: 1

Explanation

Optical character recognition, or OCR, extracts machine-readable text from images and scanned documents. It is commonly used as an initial step in document-processing pipelines when the source content is visual rather than already available as digital text. Once text has been extracted, other services can analyze, classify, summarize, or index the content. Text-to-speech converts text into audio, sentiment scoring analyzes emotional tone, and model fine-tuning changes model behavior through additional training. Therefore, OCR is the appropriate capability when the immediate requirement is to obtain text from scanned images.

Question 18

What is a key reason to evaluate a generative AI model before production deployment?

1. To remove all cloud costs
2. To measure response quality
3. To disable monitoring
4. To eliminate prompts

Correct Answer: 2

Explanation

Model evaluation helps determine whether a generative AI solution meets defined quality and performance requirements before it is released to users. Evaluation can examine factors such as response quality, relevance, groundedness, safety, consistency, and other task-specific criteria. This provides evidence that can be used to compare configurations and identify areas requiring improvement. Evaluation does not eliminate cloud costs, remove the need for monitoring, or eliminate prompts. In a production-oriented AI lifecycle, systematic evaluation supports better decisions about model selection, prompt design, retrieval configuration, and overall solution readiness.

Question 19

A company wants to process a user’s voice command and respond with spoken output. Which sequence is most appropriate?

1. OCR then search
2. Speech-to-text then text-to-speech
3. Image analysis then OCR
4. Translation then OCR

Correct Answer: 2

Explanation

A voice-based conversational application can use speech-to-text to convert the user’s spoken command into text. The application can then process that text using an AI model or another service and use text-to-speech to convert the resulting response into spoken audio. This creates a complete voice interaction flow. OCR is intended for extracting text from images, while image analysis handles visual content. Translation can be added when language conversion is required, but it does not replace the fundamental speech input and speech output stages. Therefore, speech-to-text followed by text-to-speech is the appropriate sequence.

Question 20

An Azure AI solution must be monitored for model performance and resource consumption after deployment. What should the engineer implement?

1. Diagnostic and monitoring configuration
2. Document scanning
3. Static image storage
4. Manual prompt deletion

Correct Answer: 1

Explanation

Monitoring and diagnostic configuration helps engineers observe an AI solution after deployment. Depending on the service and architecture, useful information can include performance behavior, resource consumption, errors, operational activity, and other telemetry. Monitoring is important because an AI solution can behave differently under real workloads than during development and testing. Observability can help identify performance degradation, unexpected resource usage, or operational problems that require investigation. Document scanning, image storage, and prompt deletion do not provide the monitoring capabilities needed for an operational AI solution. Effective monitoring should therefore be planned as part of deployment and ongoing management.