{"id":13766,"date":"2026-09-16T10:45:20","date_gmt":"2026-09-16T10:45:20","guid":{"rendered":"https:\/\/www.examlabs.com\/certification\/?p=13766"},"modified":"2026-09-16T10:45:20","modified_gmt":"2026-09-16T10:45:20","slug":"microsoft-ai-103-practice-test-questions-and-exam-dumps-part19-q361-380","status":"publish","type":"post","link":"https:\/\/www.examlabs.com\/certification\/microsoft-ai-103-practice-test-questions-and-exam-dumps-part19-q361-380\/","title":{"rendered":"Microsoft AI-103 Practice Test Questions and Exam Dumps Part19 Q361-380"},"content":{"rendered":"

<\/h1>\n

View Full <\/b>Microsoft AI-103 Exam Dumps<\/b><\/a> and Practice Test Dumps.<\/b><\/h2>\n

 <\/p>\n

Question 361<\/b><\/h3>\n

Which capability helps an AI agent remember information during a conversation?<\/b><\/p>\n

    \n
  1. OCR<\/span><\/li>\n
  2. Speech translation<\/span><\/li>\n
  3. State or memory management<\/span><\/li>\n
  4. Object detection<\/span><\/li>\n<\/ol>\n

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

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

    State or memory management allows an AI agent to maintain relevant information across multiple interactions. For example, an agent may need to remember a user’s selected product, previous request, or conversation context before responding to the next message. Developers should decide what information needs to be retained and for how long. Sensitive information should not be stored unnecessarily, and access controls should protect persistent state. Memory management should also account for conversation limits, stale information, and mechanisms for clearing or updating stored state.<\/span><\/p>\n

    Question 362<\/b><\/h3>\n

    What is the purpose of an AI model deployment?<\/b><\/p>\n

      \n
    1. Make a model available for application use<\/span><\/li>\n
    2. Extract text from images<\/span><\/li>\n
    3. Store search metadata<\/span><\/li>\n
    4. Detect sentiment<\/span><\/li>\n<\/ol>\n

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

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

      A model deployment makes a selected AI model available to an application through a supported endpoint or runtime configuration. Applications can then send requests to the deployment according to its capabilities and limits. Developers should consider deployment configuration, authentication, quotas, model version, capacity, latency, and cost. Testing should be performed before production use to verify that the deployed model behaves as expected. A deployment does not automatically guarantee application reliability, so monitoring, evaluation, error handling, and appropriate security controls are still required.<\/span><\/p>\n

      Question 363<\/b><\/h3>\n

      Why is reranking useful after an initial search?<\/b><\/p>\n

        \n
      1. It converts documents into audio<\/span><\/li>\n
      2. It improves the ordering of retrieved results<\/span><\/li>\n
      3. It authenticates users<\/span><\/li>\n
      4. It creates embeddings<\/span><\/li>\n<\/ol>\n

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

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

        Reranking improves the order of an initial set of retrieved documents by evaluating their relevance to the user’s query. A first-stage search can quickly identify candidate documents, while reranking can place the most useful results near the top. This is particularly valuable in RAG applications where only a limited amount of context may be supplied to the model. Developers should evaluate reranking using representative queries and consider its additional processing cost and latency. Reranking cannot compensate for completely missing or inaccessible source information.<\/span><\/p>\n

        Question 364<\/b><\/h3>\n

        Which service is designed to analyze invoices and receipts?<\/b><\/p>\n

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

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

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

          Azure AI Document Intelligence provides document-analysis capabilities that can extract information from supported business documents such as invoices and receipts. Depending on the selected model, applications can retrieve fields, tables, text, and other document information. This can reduce manual data-entry requirements in business workflows. Developers should validate extracted values because document layouts and image quality can vary. Testing should include representative documents, unusual layouts, and potential extraction errors. Appropriate access controls should also protect sensitive financial or customer information processed by the application.<\/span><\/p>\n

          Question 365<\/b><\/h3>\n

          What does grounding provide to a generative AI model?<\/b><\/p>\n

            \n
          1. External context supporting the response<\/span><\/li>\n
          2. A larger display resolution<\/span><\/li>\n
          3. Automatic administrator privileges<\/span><\/li>\n
          4. Audio compression<\/span><\/li>\n<\/ol>\n

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

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

            Grounding provides a model with relevant external information that can support its generated response. In enterprise applications, grounding may come from approved documents, databases, search results, or other trusted sources. RAG is a common approach for retrieving this information at query time. Grounding can help reduce unsupported responses, but developers should still evaluate the retrieved content and generated answer. Source freshness, relevance, authorization, and citation accuracy are important factors. Grounding should complement, rather than replace, application security and validation controls.<\/span><\/p>\n

            Question 366<\/b><\/h3>\n

            Which search method uses numerical vector representations?<\/b><\/p>\n

              \n
            1. Keyword search<\/span><\/li>\n
            2. Vector search<\/span><\/li>\n
            3. OCR<\/span><\/li>\n
            4. Sentiment analysis<\/span><\/li>\n<\/ol>\n

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

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

              Vector search uses numerical vector representations, usually created through embeddings, to identify content that is semantically similar to a query. Unlike traditional keyword search, vector search can find relevant content even when the wording differs between the query and stored documents. It is widely used in RAG and knowledge-retrieval applications. Developers should choose an appropriate embedding model and evaluate search quality with realistic queries. Metadata filters and authorization controls may also be required when the vector index contains information belonging to different users or groups.<\/span><\/p>\n

              Question 367<\/b><\/h3>\n

              What should happen when a tool receives an invalid argument?<\/b><\/p>\n

                \n
              1. Execute it anyway<\/span><\/li>\n
              2. Ignore authorization<\/span><\/li>\n
              3. Reject or validate the request<\/span><\/li>\n
              4. Increase model temperature<\/span><\/li>\n<\/ol>\n

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

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

                Tool arguments generated by an AI model should be validated before the application executes the requested operation. If an argument is missing, malformed, outside an allowed range, or inconsistent with business rules, the application should reject it or request a corrected value. This prevents accidental or malicious operations from reaching connected systems. Validation should occur outside the model because model-generated output cannot be assumed to be trustworthy. Developers should also apply authentication, authorization, logging, and least-privilege controls to sensitive tools.<\/span><\/p>\n

                Question 368<\/b><\/h3>\n

                Which capability can translate written content between supported languages?<\/b><\/p>\n

                  \n
                1. OCR<\/span><\/li>\n
                2. Machine translation<\/span><\/li>\n
                3. Object detection<\/span><\/li>\n
                4. Vector indexing<\/span><\/li>\n<\/ol>\n

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

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

                  Machine translation converts written content from one supported language into another. It can help applications provide multilingual interfaces, translate customer messages, localize content, and process information written in different languages. Developers should verify that the required language pair and features are supported for their scenario. Translation quality can vary with domain-specific terminology, context, and ambiguous expressions. Important business or legal content may require additional review. Machine translation should also be distinguished from language detection, which identifies the language of existing text rather than translating it.<\/span><\/p>\n

                  Question 369<\/b><\/h3>\n

                  Why are representative evaluation datasets important?<\/b><\/p>\n

                    \n
                  1. They reflect realistic application inputs<\/span><\/li>\n
                  2. They automatically secure databases<\/span><\/li>\n
                  3. They increase network bandwidth<\/span><\/li>\n
                  4. They eliminate monitoring<\/span><\/li>\n<\/ol>\n

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

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

                    Representative evaluation datasets contain examples that resemble the inputs an AI application is expected to receive in real use. They help developers measure performance consistently across different models, prompts, retrieval strategies, or application versions. A useful dataset should include common cases as well as important difficult or edge cases. Evaluation results can reveal regressions that may not be visible during casual testing. Developers should periodically update evaluation data when the application’s users, content, requirements, or supported scenarios change.<\/span><\/p>\n

                    Question 370<\/b><\/h3>\n

                    What is a key purpose of agent instructions?<\/b><\/p>\n

                      \n
                    1. Define the agent’s intended behavior<\/span><\/li>\n
                    2. Store search vectors<\/span><\/li>\n
                    3. Convert images into text<\/span><\/li>\n
                    4. Encrypt API traffic<\/span><\/li>\n<\/ol>\n

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

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

                      Agent instructions define how an AI agent should behave when handling requests. They can specify the agent’s role, goals, response style, limitations, and rules for using available tools. Clear instructions can make agent behavior more consistent and help establish appropriate boundaries. However, instructions alone are not a security mechanism. Developers should enforce authentication, authorization, tool restrictions, input validation, and other controls in application code. Instructions should also be tested against adversarial prompts and unexpected situations to identify weaknesses in the agent’s behavior.<\/span><\/p>\n

                      Question 371<\/b><\/h3>\n

                      Which Azure service provides speech-to-text capabilities?<\/b><\/p>\n

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

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

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

                        Azure AI Speech provides speech-to-text capabilities that convert spoken audio into written text. This can support transcription, voice-controlled applications, meeting processing, accessibility solutions, and conversational systems. Developers should consider factors such as language support, pronunciation, background noise, microphone quality, and speaker conditions when evaluating transcription quality. Real-world audio should be included in testing because ideal recordings may not represent actual usage. Speech-to-text can also be combined with language-processing or generative AI services to analyze and summarize transcribed content.<\/span><\/p>\n

                        Question 372<\/b><\/h3>\n

                        What is the main purpose of semantic ranking?<\/b><\/p>\n

                          \n
                        1. Store application secrets<\/span><\/li>\n
                        2. Order results based on semantic relevance<\/span><\/li>\n
                        3. Convert text into speech<\/span><\/li>\n
                        4. Extract invoice fields<\/span><\/li>\n<\/ol>\n

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

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

                          Semantic ranking orders search results according to their meaning and relevance to a user’s query. It can help identify documents that are conceptually related even when the exact wording differs. This can be particularly useful for natural-language questions in enterprise search and RAG applications. Semantic ranking is applied after or alongside retrieval depending on the search architecture. Developers should evaluate ranking quality with realistic queries and relevant documents. Ranking does not determine whether a user is authorized to access a document, so access controls remain necessary.<\/span><\/p>\n

                          Question 373<\/b><\/h3>\n

                          Which practice protects an application from exposing unnecessary personal data?<\/b><\/p>\n

                            \n
                          1. Data minimization<\/span><\/li>\n
                          2. Increasing temperature<\/span><\/li>\n
                          3. Disabling logging completely<\/span><\/li>\n
                          4. Removing authentication<\/span><\/li>\n<\/ol>\n

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

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

                            Data minimization means collecting, processing, and retaining only the information necessary for the intended task. In AI applications, this can reduce privacy risks by preventing unnecessary personal or confidential information from being included in prompts, logs, indexes, or application storage. Developers should determine what data is actually required before sending it to AI services. Appropriate retention, access controls, encryption, and logging policies should also be applied. Data minimization is an important privacy practice, but it should be combined with broader security and governance controls.<\/span><\/p>\n

                            Question 374<\/b><\/h3>\n

                            Which technique combines keyword and semantic retrieval signals?<\/b><\/p>\n

                              \n
                            1. OCR<\/span><\/li>\n
                            2. Hybrid search<\/span><\/li>\n
                            3. Text-to-speech<\/span><\/li>\n
                            4. Entity recognition<\/span><\/li>\n<\/ol>\n

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

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

                              Hybrid search combines keyword-based retrieval with semantic or vector-based retrieval to improve the ability to find relevant content. Keyword retrieval can be valuable for exact identifiers, names, and technical terms, while semantic retrieval can identify conceptually related information. Combining these approaches can support a broader range of user queries. Developers should evaluate the ranking configuration with realistic search examples and monitor retrieval quality. Hybrid search does not replace authorization or filtering, so applications must still ensure that returned documents are appropriate for the requesting user.<\/span><\/p>\n

                              Question 375<\/b><\/h3>\n

                              Why should an AI application maintain telemetry?<\/b><\/p>\n

                                \n
                              1. To observe performance and operational behavior<\/span><\/li>\n
                              2. To replace authentication<\/span><\/li>\n
                              3. To increase image size<\/span><\/li>\n
                              4. To create training data automatically<\/span><\/li>\n<\/ol>\n

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

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

                                Telemetry provides information about how an AI application behaves during operation. Useful telemetry can include request counts, latency, errors, token usage, retrieval performance, tool failures, and other application-specific signals. This information helps developers identify reliability problems, unexpected usage, and performance changes. Telemetry should be designed carefully so that sensitive information is not unnecessarily captured in logs. Developers should define appropriate retention and access controls. Effective monitoring combines operational telemetry with quality evaluation so both system health and AI behavior can be observed.<\/span><\/p>\n

                                Question 376<\/b><\/h3>\n

                                What is the purpose of a search indexer?<\/b><\/p>\n

                                  \n
                                1. Automatically process data sources into an index<\/span><\/li>\n
                                2. Convert speech into audio<\/span><\/li>\n
                                3. Assign user permissions<\/span><\/li>\n
                                4. Generate model temperature values<\/span><\/li>\n<\/ol>\n

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

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

                                  A search indexer can automate the process of reading supported data from a configured source, extracting relevant content, and populating a search index. This can simplify ingestion for document repositories and other supported data sources. Developers should configure field mappings, schedules, data-change handling, and enrichment requirements according to the application. Indexing quality directly affects retrieval quality, so failures and outdated content should be monitored. Security requirements must also be considered to ensure that indexed information is not exposed to users who are not authorized to access it.<\/span><\/p>\n

                                  Question 377<\/b><\/h3>\n

                                  Which capability detects whether text contains a person’s name or organization?<\/b><\/p>\n

                                    \n
                                  1. Speech synthesis<\/span><\/li>\n
                                  2. Semantic ranking<\/span><\/li>\n
                                  3. Named entity recognition<\/span><\/li>\n
                                  4. Vector search<\/span><\/li>\n<\/ol>\n

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

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

                                    Named entity recognition identifies specific entities in text, including supported categories such as people, organizations, locations, dates, and other named items. This capability can help applications extract structured information from unstructured text. For example, a document-processing solution can identify organization names and locations for indexing or analysis. Developers should evaluate the results using representative domain-specific content because unusual terminology, abbreviations, and ambiguous expressions can affect recognition. Entity extraction should also be validated before it is used to trigger important automated actions.<\/span><\/p>\n

                                    Question 378<\/b><\/h3>\n

                                    What is the purpose of a model fallback strategy?<\/b><\/p>\n

                                      \n
                                    1. Provide an alternative when the primary model is unavailable or unsuitable<\/span><\/li>\n
                                    2. Increase document resolution<\/span><\/li>\n
                                    3. Remove all application logs<\/span><\/li>\n
                                    4. Disable access controls<\/span><\/li>\n<\/ol>\n

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

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

                                      A model fallback strategy provides an alternative processing path when the primary model cannot handle a request or becomes unavailable. A fallback might use another supported model or a simpler processing approach, depending on the application. Developers should define when fallback occurs and verify that the alternative meets required quality, latency, cost, and safety requirements. Fallback logic should not bypass security or validation controls. Applications should also monitor fallback frequency because frequent fallback events may indicate capacity, availability, configuration, or workload problems.<\/span><\/p>\n

                                      Question 379<\/b><\/h3>\n

                                      Which approach can help reduce irrelevant information in a RAG prompt?<\/b><\/p>\n

                                        \n
                                      1. Retrieve and rank only relevant context<\/span><\/li>\n
                                      2. Add every indexed document<\/span><\/li>\n
                                      3. Disable search filters<\/span><\/li>\n
                                      4. Increase document duplication<\/span><\/li>\n<\/ol>\n

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

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

                                        Retrieving and ranking relevant context helps limit the information supplied to a generative model. Sending too many unrelated documents can consume tokens and may make it harder for the model to identify the information needed to answer the user’s question. Developers can improve context quality through appropriate chunking, metadata filtering, semantic or vector search, hybrid retrieval, and reranking. The selected context should also respect authorization requirements. Retrieval evaluation is important because reducing context size is useful only when the remaining information still contains the evidence needed for a correct response.<\/span><\/p>\n

                                        Question 380<\/b><\/h3>\n

                                        What should developers do before releasing an AI agent to production?<\/b><\/p>\n

                                          \n
                                        1. Skip security testing<\/span><\/li>\n
                                        2. Test tools, permissions, behavior, and edge cases<\/span><\/li>\n
                                        3. Disable monitoring<\/span><\/li>\n
                                        4. Allow unrestricted tool access<\/span><\/li>\n<\/ol>\n

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

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

                                          Before releasing an AI agent to production, developers should test its instructions, connected tools, permissions, input handling, output validation, and behavior under realistic and adversarial scenarios. Security testing should include prompt injection attempts, unauthorized access attempts, invalid tool arguments, and unexpected inputs. Evaluation should also cover response quality, latency, reliability, and important edge cases. Monitoring should be enabled before release so problems can be detected after deployment. Production readiness requires multiple layers of controls rather than relying only on the model’s instructions.<\/span><\/p>\n

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

                                          View Full Microsoft AI-103 Exam Dumps and Practice Test Dumps.   Question 361 Which capability helps an AI agent remember information during a conversation? OCR Speech translation State or memory management Object detection Correct Answer: 4 Explanation State or memory management allows an AI agent to maintain relevant information across multiple interactions. For example, an […]<\/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\/13766"}],"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=13766"}],"version-history":[{"count":1,"href":"https:\/\/www.examlabs.com\/certification\/wp-json\/wp\/v2\/posts\/13766\/revisions"}],"predecessor-version":[{"id":13769,"href":"https:\/\/www.examlabs.com\/certification\/wp-json\/wp\/v2\/posts\/13766\/revisions\/13769"}],"wp:attachment":[{"href":"https:\/\/www.examlabs.com\/certification\/wp-json\/wp\/v2\/media?parent=13766"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.examlabs.com\/certification\/wp-json\/wp\/v2\/categories?post=13766"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.examlabs.com\/certification\/wp-json\/wp\/v2\/tags?post=13766"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}