View Full Microsoft AI-900 Exam Dumps and Practice Test Dumps.

 

Question 41

A company wants to identify names of people, organizations, and locations mentioned in customer emails. Which natural language processing capability is most appropriate?

1. Sentiment analysis
2. Key phrase extraction
3. Named entity recognition
4. Text translation

Correct Answer: 3

Explanation

Named entity recognition identifies specific types of entities mentioned in text, such as people, organizations, locations, dates, and other recognized categories. For example, an email mentioning a customer’s name, a company, and a city could be analyzed to identify each of these entities separately. Sentiment analysis focuses on emotional tone, key phrase extraction identifies important terms or concepts, and translation converts text between languages. Named entity recognition is therefore the most suitable capability when an application needs to locate and categorize specific entities within unstructured text.

Question 42

Which AI capability converts written text into spoken audio?

1. Speech synthesis
2. Speech recognition
3. OCR
4. Sentiment analysis

Correct Answer: 1

Explanation

Speech synthesis, also known as text-to-speech, converts written text into spoken audio. It allows applications to produce a voice output from text, making it useful for virtual assistants, accessibility solutions, navigation systems, and applications that need to read information aloud. Speech recognition performs the opposite general operation by converting spoken language into text. OCR extracts written or printed text from images, while sentiment analysis evaluates the emotional tone of text. Therefore, speech synthesis is the appropriate capability when the required output is spoken audio generated from written content.

Question 43

A business receives scanned forms and wants to extract the printed words from the images so the information can be processed digitally. Which capability should it use?

1. Clustering
2. Optical character recognition
3. Regression
4. Speech synthesis

Correct Answer: 2

Explanation

Optical character recognition, or OCR, is used to recognize text contained in images or scanned documents and convert it into machine-readable text. This allows organizations to process information that originally exists in a visual format. OCR can be used as part of document processing workflows where extracted text is subsequently analyzed, searched, or stored. Clustering groups similar data, regression predicts numerical values, and speech synthesis creates spoken audio from text. Since the business needs to extract printed words from scanned forms, OCR is the appropriate capability.

Question 44

A security application needs to determine whether an uploaded image contains a dog, cat, car, or person, with one primary category assigned to each image. Which computer vision task is most suitable?

1. Object detection
2. OCR
3. Image classification
4. Speech recognition

Correct Answer: 3

Explanation

Image classification assigns one or more predefined categories to an image based on its visual content. In this scenario, the application needs to determine the primary category represented by each uploaded image, such as dog, cat, car, or person. Object detection would be more appropriate if the system needed to identify multiple objects and determine their locations within the image. OCR focuses on extracting text, while speech recognition processes spoken language. Therefore, image classification is the most suitable computer vision task for assigning categories to complete images.

Question 45

An organization is designing an AI-powered application for users with different abilities, languages, and backgrounds. Which responsible AI principle should guide this requirement?

1. Fairness
2. Inclusiveness
3. Transparency
4. Reliability

Correct Answer: 2

Explanation

Inclusiveness means designing AI systems so that they can be used effectively by people with different abilities, characteristics, languages, and backgrounds. An inclusive system should consider the needs of the intended user population rather than assuming that every user interacts with technology in the same way. Fairness focuses on equitable treatment and outcomes, transparency concerns understandable AI behavior, and reliability focuses on dependable operation. When an organization is specifically considering diverse users and accessibility requirements during AI solution design, inclusiveness is the most relevant responsible AI principle.

Question 46

Who should ultimately be responsible for decisions made using an AI system in an organization?

1. The AI model itself
2. The training dataset
3. The algorithm’s prediction
4. Appropriate human or organizational stakeholders

Correct Answer: 4

Explanation

AI systems do not independently assume organizational accountability for the decisions they support. Appropriate people and organizations remain responsible for defining how AI is used, monitoring its behavior, addressing risks, and making decisions about its deployment. Human oversight is especially important when AI outputs can significantly affect people or business operations. A model, dataset, or prediction is a technical component rather than an accountable stakeholder. Establishing clear responsibility helps organizations manage AI systems appropriately and respond when errors, unexpected behavior, or other risks occur.

Question 47

Which Azure service provides natural language capabilities such as sentiment analysis, entity recognition, and key phrase extraction?

1. Azure AI Language
2. Azure AI Vision
3. Azure AI Speech
4. Azure Machine Learning

Correct Answer: 1

Explanation

Azure AI Language provides a range of natural language processing capabilities for analyzing and understanding text. These capabilities include sentiment analysis, named entity recognition, key phrase extraction, language-related analysis, and other language processing features. Azure AI Vision focuses on images and visual information, while Azure AI Speech provides speech-related capabilities such as speech recognition and synthesis. Azure Machine Learning provides a broader platform for developing and managing machine learning models. Therefore, Azure AI Language is the appropriate service for the listed text analysis requirements.

Question 48

A company wants to create an application that can understand customer questions and maintain a conversation with users. Which type of AI solution is most appropriate?

1. Computer vision
2. Conversational AI
3. Regression
4. Anomaly detection

Correct Answer: 2

Explanation

Conversational AI enables applications to interact with users through natural language in a conversational manner. Such systems can understand questions, identify user intent, retrieve relevant information, and generate or select appropriate responses. They can be implemented through combinations of language processing, knowledge sources, generative AI, and other components. Computer vision processes visual information, regression predicts numerical values, and anomaly detection identifies unusual patterns. Since the company needs an application capable of understanding customer questions and participating in conversations, conversational AI is the appropriate solution type.

Question 49

A marketing team wants to automatically identify important concepts such as “customer service,” “delivery delay,” and “product quality” from thousands of reviews. Which capability should it use?

1. Speech recognition
2. Object detection
3. Key phrase extraction
4. Image classification

Correct Answer: 3

Explanation

Key phrase extraction identifies important words and phrases that represent the main concepts within a piece of text. For customer reviews, it can help surface phrases such as product quality, delivery delay, or customer service without requiring the organization to manually read every review. This capability can support document analysis, customer feedback processing, and content summarization workflows. Speech recognition works with spoken audio, object detection identifies objects in images, and image classification assigns categories to images. Key phrase extraction is therefore the best match for identifying important concepts in written reviews.

Question 50

A global company needs to convert customer support messages written in Spanish into English so that its English-speaking support team can review them. Which AI capability is required?

1. Sentiment analysis
2. Entity recognition
3. Translation
4. OCR

Correct Answer: 3

Explanation

Translation converts text from one language into another while attempting to preserve its meaning. In this scenario, Spanish customer support messages need to be converted into English so that an English-speaking team can understand and process them. Translation capabilities can support multilingual communication across applications and business workflows. Sentiment analysis determines emotional tone, entity recognition identifies specific entities in text, and OCR extracts text from images or scanned documents. Since the requirement involves converting written content from Spanish into English, translation is the appropriate AI capability.

Question 51

A smart home device needs to recognize spoken commands such as “turn on the lights” and convert them into text for further processing. Which capability should be used?

1. Speech-to-text
2. Text-to-speech
3. Image classification
4. Key phrase extraction

Correct Answer: 1

Explanation

Speech-to-text converts spoken language into written text that software applications can process. A smart home device can use this capability to transcribe commands such as “turn on the lights” before another component determines the user’s intent and performs the requested action. Text-to-speech works in the opposite direction by generating spoken audio from text. Image classification analyzes images, while key phrase extraction identifies important terms from text that already exists. Therefore, speech-to-text is the required capability for converting the user’s spoken command into machine-readable text.

Question 52

A warehouse camera must locate every package on a conveyor belt and draw a boundary around each package. Which computer vision capability is most appropriate?

1. Image classification
2. Object detection
3. Sentiment analysis
4. Language detection

Correct Answer: 2

Explanation

Object detection identifies individual objects in an image and determines their locations, commonly by using bounding boxes. For a warehouse camera, object detection can identify each package on a conveyor belt and indicate where each package appears. Image classification generally determines what category an entire image represents rather than locating multiple individual objects. Sentiment analysis and language detection are natural language capabilities and do not analyze physical objects in photographs. Because the warehouse system must locate multiple packages and outline their positions, object detection is the appropriate computer vision capability.

Question 53

An AI model performs significantly better for one group of users than another group, even though both groups are part of the intended population. Which responsible AI concern should be investigated?

1. Fairness
2. Speech synthesis
3. Data compression
4. Image resolution

Correct Answer: 1

Explanation

Fairness should be investigated when an AI system produces substantially different outcomes or performance across groups that are part of its intended population. Organizations can evaluate model behavior across relevant groups and examine differences in measures such as error rates, accuracy, or other outcomes. Investigating the training data and model behavior can help identify potential causes of disparities. Speech synthesis, data compression, and image resolution do not address equitable model performance. Fairness is therefore the responsible AI principle most directly related to differences in model outcomes across user groups.

Question 54

A user enters a text prompt describing a mountain landscape, and an AI system creates a new image based on that description. What type of AI capability is being demonstrated?

1. Classification
2. Regression
3. Generative AI
4. Clustering

Correct Answer: 3

Explanation

Generative AI can create new content based on instructions or prompts supplied by users. In this scenario, the system receives a textual description of a mountain landscape and generates a new image that reflects the description. This is an example of text-to-image generation, which is a generative AI capability. Classification assigns categories, regression predicts numerical values, and clustering discovers groups in data. The defining characteristic here is that the system produces new visual content rather than simply categorizing or predicting an existing value.

Question 55

Which statement best describes a generative AI model?

1. It can only classify existing images
2. It creates new content based on learned patterns and provided input
3. It can only calculate numerical averages
4. It must always use manually written rules for every output

Correct Answer: 2

Explanation

Generative AI models are designed to produce new content by using patterns learned from data and information provided through an interaction. Depending on the model, generated content can include text, images, audio, video, or code. A prompt or other input helps guide what the model generates. Generative AI is different from systems designed only to classify existing information or calculate fixed numerical results. Although generated content can be useful, its outputs still need appropriate evaluation because generative models can produce inaccurate or unsupported information.

Question 56

A user tells a generative AI model, “Summarize this report in five bullet points for a business executive.” What is this user-provided instruction called?

1. A prompt
2. A label
3. A feature
4. A validation dataset

Correct Answer: 1

Explanation

A prompt is an instruction or input provided to a generative AI model to guide the content or form of its response. In this example, the user specifies both the task, summarizing the report, and the desired format, five bullet points for a business executive. Clear prompts can provide context, constraints, and instructions that help guide the model toward a useful response. A label is associated with supervised learning, a feature is an input characteristic, and a validation dataset is used during machine learning development.

Question 57

A generative AI application retrieves information from a company’s approved knowledge base and includes that information in the context given to the model before generating a response. What is the main benefit of this approach?

1. It removes the need for any user input
2. It guarantees that every generated response is correct
3. It helps ground responses in relevant external information
4. It converts speech into images

Correct Answer: 3

Explanation

Grounding provides a generative AI model with relevant information from an external source so that its response can be based on information that is available and appropriate for the application. A company might retrieve content from an approved knowledge base, documents, or another trusted source and provide it as context. This can improve relevance and help reduce unsupported responses, although grounding does not guarantee that every generated answer will be completely correct. The approach is therefore useful for connecting generated responses with relevant organizational information.

Question 58

A generative AI chatbot produces an answer containing a convincing but completely unsupported fact. What limitation does this demonstrate?

1. Hallucination
2. Clustering
3. OCR
4. Regression

Correct Answer: 1

Explanation

A hallucination occurs when a generative AI system produces information that is inaccurate, fabricated, or unsupported while presenting it as though it were valid. Because generative models are designed to produce plausible responses, an output can sound confident even when the information is incorrect. Applications should therefore evaluate generated content and use appropriate grounding, instructions, safeguards, and human review where necessary. Clustering groups similar data, OCR extracts text from images, and regression predicts numerical values. None of those concepts describes unsupported generated information.

Question 59

A factory uses cameras to inspect manufactured components and determine whether each component has a visible defect. Which Azure AI workload is most directly involved?

1. Natural language processing
2. Computer vision
3. Speech recognition
4. Language translation

Correct Answer: 2

Explanation

Computer vision enables applications to analyze images and other visual information. In a manufacturing environment, cameras can capture component images and AI models can analyze those images to identify defects, classify products, or detect objects. This can support automated quality inspection and reduce the need for every item to be manually inspected. Natural language processing works with human language, speech recognition processes spoken audio, and translation converts content between languages. Because the factory’s requirement involves analyzing photographs of physical components, computer vision is the appropriate AI workload.

Question 60

A company evaluates an AI model using data that closely represents the people and conditions where the model will be used. Why is this approach important?

1. It eliminates the need for model monitoring
2. It guarantees that the model will never make an error
3. It makes every AI model identical
4. It provides a more meaningful indication of expected real-world performance

Correct Answer: 4

Explanation

Using representative evaluation data helps an organization understand how an AI model may perform in the environment and population where it is intended to operate. If evaluation data differs substantially from real-world conditions, important weaknesses may remain hidden until deployment. Representative data can include relevant variations in users, inputs, environments, and other factors that affect model performance. It does not guarantee perfect results, but it provides a more meaningful assessment of expected behavior and can help identify areas that require improvement before deployment.