{"id":18258,"date":"2026-09-22T06:22:29","date_gmt":"2026-09-22T06:22:29","guid":{"rendered":"https:\/\/www.examlabs.com\/certification\/?p=18258"},"modified":"2026-09-22T06:22:29","modified_gmt":"2026-09-22T06:22:29","slug":"microsoft-ai-900-practice-test-questions-and-exam-dumps-part8-q141-160","status":"publish","type":"post","link":"https:\/\/www.examlabs.com\/certification\/microsoft-ai-900-practice-test-questions-and-exam-dumps-part8-q141-160\/","title":{"rendered":"Microsoft AI-900 Practice Test Questions and Exam Dumps Part8 Q141-160"},"content":{"rendered":"

View Full <\/b>Microsoft AI-900 Exam Dumps<\/b><\/a> and Practice Test Dumps.<\/b><\/h2>\n

 <\/p>\n

Question 141<\/b><\/h3>\n

Which Azure service capability is designed to extract structured information from receipts, invoices, and other business documents?<\/b><\/p>\n

    \n
  1. Azure AI Speech<\/span><\/li>\n
  2. Azure AI Vision<\/span><\/li>\n
  3. Azure AI Language<\/span><\/li>\n
  4. Azure AI Document Intelligence<\/span><\/li>\n<\/ol>\n

    Correct Answer: 4<\/b><\/p>\n

    Explanation<\/b><\/p>\n

    Azure AI Document Intelligence can analyze documents and extract structured information from their content. It provides prebuilt models for common document types, including invoices and receipts, and also supports custom models for specialized documents. Extracted fields can then be used by business applications without requiring users to manually enter the information. Azure AI Speech focuses on audio and spoken language, Azure AI Vision focuses on visual analysis, and Azure AI Language handles text-based language workloads. Document Intelligence is therefore the appropriate service for structured document extraction.<\/span><\/p>\n

    Question 142<\/b><\/h3>\n

    A model predicts the number of products a customer will purchase next month. What type of machine learning task is this?<\/b><\/p>\n

      \n
    1. Classification<\/span><\/li>\n
    2. Regression<\/span><\/li>\n
    3. Clustering<\/span><\/li>\n
    4. Object detection<\/span><\/li>\n<\/ol>\n

      Correct Answer: 2<\/b><\/p>\n

      Explanation<\/b><\/p>\n

      Regression is used when a machine learning model predicts a numerical value. In this scenario, the number of products a customer is expected to purchase is a quantitative output, making regression appropriate. The model can use features such as previous purchases, customer activity, and purchase frequency to estimate the future value. Classification predicts categories, clustering identifies groups within unlabeled data, and object detection is used to locate objects in images. Because the target is a numerical quantity, the problem is best represented as regression.<\/span><\/p>\n

      Question 143<\/b><\/h3>\n

      A company groups customers according to similarities in their purchasing behavior without having predefined customer categories. Which technique is appropriate?<\/b><\/p>\n

        \n
      1. Clustering<\/span><\/li>\n
      2. Regression<\/span><\/li>\n
      3. Classification<\/span><\/li>\n
      4. Speech recognition<\/span><\/li>\n<\/ol>\n

        Correct Answer: 1<\/b><\/p>\n

        Explanation<\/b><\/p>\n

        Clustering is an unsupervised learning technique that groups data points according to similarities without requiring predefined labels. In this scenario, customer purchasing behavior can be analyzed using characteristics such as purchase frequency, average order value, and product preferences. The algorithm can identify naturally occurring groups that may help the company understand different customer segments. Classification would require predefined categories, while regression predicts numerical values. Speech recognition is unrelated because the input data consists of customer purchasing information rather than spoken language.<\/span><\/p>\n

        Question 144<\/b><\/h3>\n

        A company wants an AI system to determine whether a product image contains a damaged item or an undamaged item. Which computer vision task is most appropriate?<\/b><\/p>\n

          \n
        1. OCR<\/span><\/li>\n
        2. Object detection<\/span><\/li>\n
        3. Image classification<\/span><\/li>\n
        4. Speech recognition<\/span><\/li>\n<\/ol>\n

          Correct Answer: 3<\/b><\/p>\n

          Explanation<\/b><\/p>\n

          Image classification assigns one or more categories to an image based on its visual content. In this scenario, the system can classify a product image as showing a damaged or undamaged item. The requirement does not specify that the system must locate the damaged region, so object detection is not necessary. OCR extracts text from images, while speech recognition processes spoken audio. Image classification is therefore appropriate when the primary goal is to assign a category to the entire image based on its contents.<\/span><\/p>\n

          Question 145<\/b><\/h3>\n

          Which feature of Azure Machine Learning can help automatically select and evaluate different machine learning algorithms and configurations?<\/b><\/p>\n

            \n
          1. Automated machine learning<\/span><\/li>\n
          2. Optical character recognition<\/span><\/li>\n
          3. Speech translation<\/span><\/li>\n
          4. Image tagging<\/span><\/li>\n<\/ol>\n

            Correct Answer: 1<\/b><\/p>\n

            Explanation<\/b><\/p>\n

            Automated machine learning, often called AutoML, helps simplify the process of training machine learning models by automatically trying different algorithms, preprocessing approaches, and model configurations. It can evaluate candidate models using selected metrics and identify suitable approaches for the specified task. This can reduce the amount of manual experimentation required during model development. OCR is used for extracting text from images, speech translation handles spoken-language translation, and image tagging describes image content. AutoML is therefore the capability relevant to automated model experimentation and selection.<\/span><\/p>\n

            Question 146<\/b><\/h3>\n

            A company needs a model to recognize whether an image contains a cat, dog, bird, or other animal category. The location of the animal is not required. Which approach should be used?<\/b><\/p>\n

              \n
            1. Object detection<\/span><\/li>\n
            2. OCR<\/span><\/li>\n
            3. Image segmentation<\/span><\/li>\n
            4. Image classification<\/span><\/li>\n<\/ol>\n

              Correct Answer: 4<\/b><\/p>\n

              Explanation<\/b><\/p>\n

              Image classification is suitable when the goal is to assign a category to an image without needing to identify the precise location of objects within it. In this example, the model can determine whether the image belongs to categories such as cat, dog, or bird. Object detection would be more appropriate if the application also needed to locate individual animals. OCR extracts text, while image segmentation identifies specific regions or pixels associated with objects. Because only the animal category is required, image classification matches the scenario.<\/span><\/p>\n

              Question 147<\/b><\/h3>\n

              A language application must identify names of people, organizations, and locations mentioned in customer messages. Which capability is required?<\/b><\/p>\n

                \n
              1. Sentiment analysis<\/span><\/li>\n
              2. Translation<\/span><\/li>\n
              3. Named entity recognition<\/span><\/li>\n
              4. Speech synthesis<\/span><\/li>\n<\/ol>\n

                Correct Answer: 3<\/b><\/p>\n

                Explanation<\/b><\/p>\n

                Named entity recognition identifies specific entities mentioned in text and can categorize them according to types such as people, organizations, locations, dates, or other supported entity categories. For example, a customer message mentioning a person’s name and a company location can be analyzed to identify those entities separately from the rest of the text. Sentiment analysis determines emotional tone, translation converts text between languages, and speech synthesis produces spoken audio. Named entity recognition is therefore the appropriate capability for extracting specific entities from written messages.<\/span><\/p>\n

                Question 148<\/b><\/h3>\n

                A developer wants an AI application to convert a written response into natural-sounding spoken audio. Which Azure AI capability should be used?<\/b><\/p>\n

                  \n
                1. Speech-to-text<\/span><\/li>\n
                2. Text-to-speech<\/span><\/li>\n
                3. OCR<\/span><\/li>\n
                4. Language detection<\/span><\/li>\n<\/ol>\n

                  Correct Answer: 2<\/b><\/p>\n

                  Explanation<\/b><\/p>\n

                  Text-to-speech converts written text into spoken audio. It can be used in applications such as virtual assistants, accessibility tools, navigation systems, and automated announcements. Speech-to-text performs the opposite direction by converting spoken language into written text. OCR extracts text from images, while language detection identifies the language used in text. When an application already has written content and needs to produce an audible response, text-to-speech is the capability that directly satisfies the requirement.<\/span><\/p>\n

                  Question 149<\/b><\/h3>\n

                  A customer asks a chatbot, “What is the return policy for electronics?” The application retrieves the relevant company policy before generating its answer. What is the main benefit of retrieving this information?<\/b><\/p>\n

                    \n
                  1. It increases image resolution<\/span><\/li>\n
                  2. It converts text into speech<\/span><\/li>\n
                  3. It provides relevant context for the generated response<\/span><\/li>\n
                  4. It automatically retrains the language model<\/span><\/li>\n<\/ol>\n

                    Correct Answer: 3<\/b><\/p>\n

                    Explanation<\/b><\/p>\n

                    Retrieving relevant information provides the generative AI model with additional context that can be used when producing an answer. In a company-policy chatbot, retrieving the appropriate return-policy content can help the response remain grounded in the organization’s available information. Retrieval does not automatically retrain the underlying language model, nor does it convert text to speech or increase image resolution. This approach is particularly useful when the required information is specific to an organization or may change over time and therefore should come from an external knowledge source.<\/span><\/p>\n

                    Question 150<\/b><\/h3>\n

                    Which responsible AI principle focuses on assigning responsibility for decisions and outcomes produced by an AI system?<\/b><\/p>\n

                      \n
                    1. Accountability<\/span><\/li>\n
                    2. Fairness<\/span><\/li>\n
                    3. Inclusiveness<\/span><\/li>\n
                    4. Transparency<\/span><\/li>\n<\/ol>\n

                      Correct Answer: 1<\/b><\/p>\n

                      Explanation<\/b><\/p>\n

                      Accountability means that appropriate people and organizations remain responsible for the development, deployment, monitoring, and outcomes of AI systems. Even when an automated system makes a recommendation or prediction, organizations should establish appropriate governance and responsibility for how the system is used. Fairness focuses on equitable treatment and outcomes, inclusiveness considers diverse users and abilities, and transparency focuses on making AI behavior understandable. Accountability is therefore the responsible AI principle most directly associated with responsibility for AI decisions and their consequences.<\/span><\/p>\n

                      Question 151<\/b><\/h3>\n

                      A company wants to translate customer messages from Spanish into English before analyzing their sentiment. Which sequence of AI capabilities would satisfy this requirement?<\/b><\/p>\n

                        \n
                      1. Speech synthesis followed by OCR<\/span><\/li>\n
                      2. Translation followed by sentiment analysis<\/span><\/li>\n
                      3. Object detection followed by classification<\/span><\/li>\n
                      4. Image tagging followed by speech recognition<\/span><\/li>\n<\/ol>\n

                        Correct Answer: 2<\/b><\/p>\n

                        Explanation<\/b><\/p>\n

                        Translation can first convert the Spanish customer messages into English, after which sentiment analysis can evaluate the emotional tone or opinion expressed in the translated text. This sequence allows the organization to perform sentiment analysis in the desired language. Speech synthesis and OCR address different types of input and output, while object detection, image tagging, and speech recognition are not required for written customer messages. The important point is that the text should first be converted into the language needed for the subsequent language analysis task.<\/span><\/p>\n

                        Question 152<\/b><\/h3>\n

                        A model designed to predict whether a transaction is fraudulent produces many legitimate transactions as fraud alerts. Which type of error is occurring frequently?<\/b><\/p>\n

                          \n
                        1. False negatives<\/span><\/li>\n
                        2. True negatives<\/span><\/li>\n
                        3. False positives<\/span><\/li>\n
                        4. True positives<\/span><\/li>\n<\/ol>\n

                          Correct Answer: 3<\/b><\/p>\n

                          Explanation<\/b><\/p>\n

                          A false positive occurs when a model predicts that a condition is present when it is actually absent. In this scenario, legitimate transactions are incorrectly classified as fraudulent, creating false fraud alerts. False negatives occur when fraudulent transactions are incorrectly classified as legitimate. True positives and true negatives represent correct predictions. Understanding these error types is important when evaluating classification models because different applications have different costs associated with false positives and false negatives. Fraud detection systems often need to balance these errors according to business requirements.<\/span><\/p>\n

                          Question 153<\/b><\/h3>\n

                          Which Azure AI Language capability can identify the overall positive, negative, or neutral opinion expressed in a customer review?<\/b><\/p>\n

                            \n
                          1. Sentiment analysis<\/span><\/li>\n
                          2. OCR<\/span><\/li>\n
                          3. Object detection<\/span><\/li>\n
                          4. Speech synthesis<\/span><\/li>\n<\/ol>\n

                            Correct Answer: 1<\/b><\/p>\n

                            Explanation<\/b><\/p>\n

                            Sentiment analysis evaluates text to determine the expressed emotional or opinion-based tone. For customer reviews, it can help identify whether the content is positive, negative, or neutral and may provide more detailed sentiment information depending on the service capabilities. OCR is used to extract text from images, object detection identifies and locates objects in images, and speech synthesis converts text into spoken audio. Since the requirement concerns the opinion expressed in written customer feedback, sentiment analysis is the directly relevant Azure AI Language capability.<\/span><\/p>\n

                            Question 154<\/b><\/h3>\n

                            An AI solution must identify the language used in a customer message before sending it to the appropriate translation workflow. Which capability should be used?<\/b><\/p>\n

                              \n
                            1. Key phrase extraction<\/span><\/li>\n
                            2. Entity recognition<\/span><\/li>\n
                            3. Language detection<\/span><\/li>\n
                            4. Sentiment analysis<\/span><\/li>\n<\/ol>\n

                              Correct Answer: 3<\/b><\/p>\n

                              Explanation<\/b><\/p>\n

                              Language detection identifies the language used in a piece of text. In a multilingual customer-support system, this capability can determine whether a message is written in English, Spanish, French, or another supported language before routing it to an appropriate translation or processing workflow. Key phrase extraction identifies important concepts, entity recognition identifies names and other entities, and sentiment analysis evaluates opinion or emotional tone. Because the requirement is specifically to determine which language a message uses, language detection is the appropriate capability.<\/span><\/p>\n

                              Question 155<\/b><\/h3>\n

                              A data scientist wants to make a model less sensitive to differences in the numerical scales of input features such as income and age. What preprocessing activity can help?<\/b><\/p>\n

                                \n
                              1. Data encryption<\/span><\/li>\n
                              2. Feature scaling<\/span><\/li>\n
                              3. Speech recognition<\/span><\/li>\n
                              4. Entity extraction<\/span><\/li>\n<\/ol>\n

                                Correct Answer: 2<\/b><\/p>\n

                                Explanation<\/b><\/p>\n

                                Feature scaling transforms numerical features so that differences in their scales do not disproportionately affect certain machine learning algorithms. For example, annual income may have values in the tens of thousands while age may typically be below one hundred. Scaling can place these features into more comparable numerical ranges. Data encryption protects information, speech recognition processes spoken language, and entity extraction identifies entities in text. Feature scaling is therefore a preprocessing activity that can help prepare numerical inputs for appropriate machine learning models.<\/span><\/p>\n

                                Question 156<\/b><\/h3>\n

                                A generative AI model produces an answer that sounds convincing but contains information that is not supported by the available source material. What issue does this illustrate?<\/b><\/p>\n

                                  \n
                                1. Classification<\/span><\/li>\n
                                2. Feature scaling<\/span><\/li>\n
                                3. Hallucination<\/span><\/li>\n
                                4. Clustering<\/span><\/li>\n<\/ol>\n

                                  Correct Answer: 3<\/b><\/p>\n

                                  Explanation<\/b><\/p>\n

                                  A hallucination occurs when a generative AI model produces information that may appear plausible but is inaccurate, unsupported, or fabricated. This can be a significant concern in applications where users expect factual answers. Techniques such as grounding responses with relevant source information, carefully designing prompts, validating outputs, and applying appropriate evaluation methods can help reduce the impact of unsupported responses. Classification, feature scaling, and clustering are machine learning concepts that do not describe this generative AI behavior. Therefore, the scenario illustrates a hallucination.<\/span><\/p>\n

                                  Question 157<\/b><\/h3>\n

                                  An organization wants to identify important terms such as “refund,” “damaged product,” and “delivery delay” from customer feedback. Which language capability is most appropriate?<\/b><\/p>\n

                                    \n
                                  1. Key phrase extraction<\/span><\/li>\n
                                  2. Speech synthesis<\/span><\/li>\n
                                  3. Object detection<\/span><\/li>\n
                                  4. Image classification<\/span><\/li>\n<\/ol>\n

                                    Correct Answer: 1<\/b><\/p>\n

                                    Explanation<\/b><\/p>\n

                                    Key phrase extraction identifies important words or phrases that represent the main concepts in a piece of text. For customer feedback, it can highlight terms such as refund, damaged product, delivery delay, or other subjects that summarize the message. This can help organizations organize feedback and identify recurring issues without manually reading every message. Speech synthesis creates spoken audio, object detection identifies objects in images, and image classification assigns categories to images. Key phrase extraction is therefore the appropriate capability for identifying significant concepts within customer feedback.<\/span><\/p>\n

                                    Question 158<\/b><\/h3>\n

                                    A model is evaluated using data that was also used extensively to train the model. Why might this produce an unreliable assessment of the model’s ability to generalize?<\/b><\/p>\n

                                      \n
                                    1. The data contains too many images<\/span><\/li>\n
                                    2. The model has already seen the examples<\/span><\/li>\n
                                    3. The model cannot perform inference<\/span><\/li>\n
                                    4. The data automatically becomes encrypted<\/span><\/li>\n<\/ol>\n

                                      Correct Answer: 2<\/b><\/p>\n

                                      Explanation<\/b><\/p>\n

                                      Evaluating a model using the same data used for training can provide an overly optimistic assessment because the model has already learned from those examples. A model may perform very well on familiar training data while performing poorly on new, unseen data. Separating training and evaluation data provides a better indication of generalization. This is especially important when detecting overfitting. The issue is not related to image quantity, encryption, or the model’s ability to perform inference. Unseen evaluation data provides a more meaningful measure of expected real-world performance.<\/span><\/p>\n

                                      Question 159<\/b><\/h3>\n

                                      A company wants to make its AI assistant respond more consistently to the same type of request. Which action can help control variation in generated responses?<\/b><\/p>\n

                                        \n
                                      1. Increase image size<\/span><\/li>\n
                                      2. Remove all instructions<\/span><\/li>\n
                                      3. Adjust the temperature setting<\/span><\/li>\n
                                      4. Replace text with OCR<\/span><\/li>\n<\/ol>\n

                                        Correct Answer: 3<\/b><\/p>\n

                                        Explanation<\/b><\/p>\n

                                        Temperature is a generative AI parameter that influences the randomness of generated responses. Lowering the temperature generally makes output more predictable and consistent, while higher values can encourage more variation and creativity. The appropriate setting depends on the application’s requirements. Increasing image size does not control language generation, removing instructions can reduce consistency, and OCR is used to extract text from images. Therefore, adjusting the temperature setting can help an application produce responses with a more controlled level of variation.<\/span><\/p>\n

                                        Question 160<\/b><\/h3>\n

                                        A company wants to determine whether its AI model is performing adequately before making it available to customers. Which activity should be performed?<\/b><\/p>\n

                                          \n
                                        1. Delete all evaluation data<\/span><\/li>\n
                                        2. Disable model monitoring<\/span><\/li>\n
                                        3. Increase the number of user accounts<\/span><\/li>\n
                                        4. Evaluate the model against appropriate performance criteria<\/span><\/li>\n<\/ol>\n

                                          Correct Answer: 4<\/b><\/p>\n

                                          Explanation<\/b><\/p>\n

                                          Model evaluation involves measuring performance against criteria that are appropriate for the specific AI task and business requirement. For a classification model, metrics such as accuracy, precision, recall, or F1 score may be relevant, while regression models can use measures such as mean absolute error. Evaluation should use suitable data that represents the expected workload. Deleting evaluation data or disabling monitoring would make assessment more difficult. Increasing user accounts does not measure model quality. Appropriate performance evaluation should therefore occur before production use.<\/span><\/p>\n

                                           <\/p>\n","protected":false},"excerpt":{"rendered":"

                                          View Full Microsoft AI-900 Exam Dumps and Practice Test Dumps.   Question 141 Which Azure service capability is designed to extract structured information from receipts, invoices, and other business documents? Azure AI Speech Azure AI Vision Azure AI Language Azure AI Document Intelligence Correct Answer: 4 Explanation Azure AI Document Intelligence can analyze documents and […]<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":[],"categories":[1648,1647],"tags":[],"_links":{"self":[{"href":"https:\/\/www.examlabs.com\/certification\/wp-json\/wp\/v2\/posts\/18258"}],"collection":[{"href":"https:\/\/www.examlabs.com\/certification\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.examlabs.com\/certification\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.examlabs.com\/certification\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.examlabs.com\/certification\/wp-json\/wp\/v2\/comments?post=18258"}],"version-history":[{"count":1,"href":"https:\/\/www.examlabs.com\/certification\/wp-json\/wp\/v2\/posts\/18258\/revisions"}],"predecessor-version":[{"id":18259,"href":"https:\/\/www.examlabs.com\/certification\/wp-json\/wp\/v2\/posts\/18258\/revisions\/18259"}],"wp:attachment":[{"href":"https:\/\/www.examlabs.com\/certification\/wp-json\/wp\/v2\/media?parent=18258"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.examlabs.com\/certification\/wp-json\/wp\/v2\/categories?post=18258"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.examlabs.com\/certification\/wp-json\/wp\/v2\/tags?post=18258"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}