McAfee Secure

Exam Code: AI-901

Exam Name: Microsoft Azure AI Fundamentals

Certification Provider: Microsoft

Corresponding Certification: Microsoft Certified: Azure AI Fundamentals

Microsoft AI-901 Questions & Answers

Study with Up-To-Date REAL Exam Questions and Answers from the ACTUAL Test

166 Questions & Answers with Testing Engine

"Microsoft Azure AI Fundamentals Exam", also known as AI-901 exam, is a Microsoft certification exam.

Pass your tests with the always up-to-date AI-901 Exam Engine. Your AI-901 training materials keep you at the head of the pack!

Satisfaction Guaranteed

Money Back Guarantee

Testking provides hassle-free money back guarantee with our products. That is because we have 100% trust in the abilities of our professional and experience product team, and our record is a proof of that.

99.6% PASS RATE
Was: $137.49
Now: $124.99

Microsoft AI-901 Practice Test 74

Question 730

An Azure AI solution uses the following approved control: Require governance evidence with reviewers, test outcomes, and documented exceptions. What observable result should pass QA?

  1. A screen-reader user can complete the service flow without mouse-only actions
  2. A legally ambiguous case offers appropriate caveats and referral
  3. A compliance record links each exception to an owner and disposition
  4. An ambiguous request results in a clarification question and no side effect

Explanation

Correct Answer

C:

Self-attestation is not independent proof of regulatory controls. The expected acceptance evidence is: A compliance record links each exception to an owner and disposition. This observation directly tests the stated behavior.

Incorrect Answers

A:

Inclusive design permits nonvisual navigation of the same task path. This observation does not establish the required result: A compliance record links each exception to an owner and disposition.

B:

Transparency reduces mistaken reliance on unsupported professional guarantees. This observation does not establish the required result: A compliance record links each exception to an owner and disposition.

D:

Reliable assistance distinguishes missing intent from authorized action. This observation does not establish the required result: A compliance record links each exception to an owner and disposition.

Question 731

A solution team has selected this control: Check regional model availability and data handling requirements before deployment. Which observation is the most direct acceptance evidence?

  1. Injection attempts fail to access protected secrets or tools
  2. The response contains only permitted fields with correct types
  3. A representative maximum-size test stays within documented budgets
  4. The target model deploys in the approved region and meets residency controls

Explanation

Correct Answer

D:

Not every model or feature is available in every location. The expected acceptance evidence is: The target model deploys in the approved region and meets residency controls. This observation directly tests the stated behavior.

Incorrect Answers

A:

A user cannot use text to elevate permissions beyond the application policy. This observation does not establish the required result: The target model deploys in the approved region and meets residency controls.

B:

Model output constraints help downstream parsing but still require validation. This observation does not establish the required result: The target model deploys in the approved region and meets residency controls.

C:

Capabilities and limits vary by model deployment. This observation does not establish the required result: The target model deploys in the approved region and meets residency controls.

Question 732

An acceptance tester needs to validate: Maintain a governed feature inventory and trace decisions to policy. Which observed result proves the intended behavior?

  1. An auditor can map each used input to an approved policy criterion
  2. An unsafe self-treatment case triggers a safe refusal or professional referral
  3. Each policy answer citation resolves to the correct approved paragraph
  4. A consented test records the notice and applicable retention choice

Explanation

Correct Answer

A:

Accountability requires interpretable reasons for consequential decisions. The expected acceptance evidence is: An auditor can map each used input to an approved policy criterion. This observation directly tests the stated behavior.

Incorrect Answers

B:

Safety mechanisms should reduce unsupported high-impact recommendations. This observation does not establish the required result: An auditor can map each used input to an approved policy criterion.

C:

Transparency lets users inspect evidence instead of accepting fluent text. This observation does not establish the required result: An auditor can map each used input to an approved policy criterion.

D:

Privacy requires meaningful notice and bounded handling of personal data. This observation does not establish the required result: An auditor can map each used input to an approved policy criterion.

Question 733

Troubleshooting report: A retrieved document is inserted as if it were a privileged system rule. Which correction is MOST directly targeted to this failure?

  1. Use a speech synthesis endpoint and compatible output format
  2. Process appropriate video frames and the audio track
  3. Split data by appropriate entities and evaluate on unseen cases
  4. Keep external content in lower-trust user/tool data fields

Explanation

Correct Answer

D:

This correction directly addresses the reported failure. Instruction roles help distinguish application policy from user task text. The targeted change is: Keep external content in lower-trust user/tool data fields.

Incorrect Answers

A:

Speech synthesis generates audio from text rather than transcribing audio. In the stated case, troubleshooting report: A retrieved document is inserted as if it were a privileged system rule; “Use a speech synthesis endpoint and compatible output format” does not target that failure.

B:

Different media channels provide complementary scene and dialogue evidence. In the stated case, troubleshooting report: A retrieved document is inserted as if it were a privileged system rule; “Process appropriate video frames and the audio track” does not target that failure.

C:

A model can memorize training examples without generalizing. In the stated case, troubleshooting report: A retrieved document is inserted as if it were a privileged system rule; “Split data by appropriate entities and evaluate on unseen cases” does not target that failure.

Question 734

Troubleshooting report: The team assumes portal screenshots prove availability everywhere. Which correction is MOST directly targeted to this failure?

  1. Rewrite the prompt with explicit acceptance criteria
  2. Move the model call behind a secured server endpoint
  3. Enforce a typed response schema and fail invalid output
  4. Validate the exact region and supported deployment type

Explanation

Correct Answer

D:

This correction directly addresses the reported failure. Not every model or feature is available in every location. The targeted change is: Validate the exact region and supported deployment type.

Incorrect Answers

A:

Specific instructions reduce ambiguity about what constitutes success. In the stated case, troubleshooting report: The team assumes portal screenshots prove availability everywhere; “Rewrite the prompt with explicit acceptance criteria” does not target that failure.

B:

Backend controls protect credentials, quotas and data policy. In the stated case, troubleshooting report: The team assumes portal screenshots prove availability everywhere; “Move the model call behind a secured server endpoint” does not target that failure.

C:

Model output constraints help downstream parsing but still require validation. In the stated case, troubleshooting report: The team assumes portal screenshots prove availability everywhere; “Enforce a typed response schema and fail invalid output” does not target that failure.

Question 735

An engineering review documents this requirement: An uploaded image is too large or has an unsupported encoding for analysis. Select the most appropriate solution.

  1. Validate file type, size and image dimensions before the model call
  2. Preserve region coordinates with extracted text or field evidence
  3. Apply supported image safety filtering and review before distribution
  4. Normalize image orientation before applying vision detection

Explanation

Correct Answer

A:

Vision endpoints impose supported input formats and limits. This implementation addresses the precise goal rather than an adjacent AI task.

Incorrect Answers

B:

Spatial provenance supports later verification of recognition results. That is not sufficient here: Vision endpoints impose supported input formats and limits.

C:

Generative media needs policy controls in addition to prompt instructions. That is not sufficient here: Vision endpoints impose supported input formats and limits.

D:

Rotated inputs can degrade detection or coordinate interpretation. That is not sufficient here: Vision endpoints impose supported input formats and limits.

Question 736

Troubleshooting report: A shared mutable message array leaks one customer prompt into another reply. Which correction is MOST directly targeted to this failure?

  1. Add the actual user content to the chat message sequence
  2. Isolate request state and bind responses to their originating session
  3. Move system instructions into protected application configuration
  4. Constrain request size and trim or chunk input before sending

Explanation

Correct Answer

B:

This correction directly addresses the reported failure. Concurrent inference must not mix user data or response ownership. The targeted change is: Isolate request state and bind responses to their originating session.

Incorrect Answers

A:

The inference service needs task content to generate a relevant answer. In the stated case, troubleshooting report: A shared mutable message array leaks one customer prompt into another reply; “Add the actual user content to the chat message sequence” does not target that failure.

C:

Versioned system messages provide a controlled behavioral baseline. In the stated case, troubleshooting report: A shared mutable message array leaks one customer prompt into another reply; “Move system instructions into protected application configuration” does not target that failure.

D:

Capabilities and limits vary by model deployment. In the stated case, troubleshooting report: A shared mutable message array leaks one customer prompt into another reply; “Constrain request size and trim or chunk input before sending” does not target that failure.

Question 737

Troubleshooting report: The service returns isolated search links without using conversation context. Which correction is MOST directly targeted to this failure?

  1. Add bounded conversation state and retrieved evidence
  2. Train or tune a contextual recommender and evaluate precision at rank
  3. Use sentiment analysis instead of named-entity extraction
  4. Use document layout extraction with typed checkbox outputs

Explanation

Correct Answer

A:

This correction directly addresses the reported failure. The assistant should incorporate context while remaining connected to reliable sources. The targeted change is: Add bounded conversation state and retrieved evidence.

Incorrect Answers

B:

Recommendations rank candidate items rather than simply generating arbitrary titles. In the stated case, troubleshooting report: The service returns isolated search links without using conversation context; “Train or tune a contextual recommender and evaluate precision at rank” does not target that failure.

C:

Sentiment analysis infers expressed attitude rather than extracting names. In the stated case, troubleshooting report: The service returns isolated search links without using conversation context; “Use sentiment analysis instead of named-entity extraction” does not target that failure.

D:

Different form elements require more than ordinary running-text OCR. In the stated case, troubleshooting report: The service returns isolated search links without using conversation context; “Use document layout extraction with typed checkbox outputs” does not target that failure.

Question 738

An acceptance tester needs to validate: Select an image-generation-capable model and a supported image endpoint. Which observed result proves the intended behavior?

  1. The response includes the requested image artifact and metadata
  2. Unsupported factual claims are flagged and answered with uncertainty
  3. The holdout report identifies missed positives and false alarms explicitly
  4. A user request cannot override the controlling system-level instructions in tests

Explanation

Correct Answer

A:

Text generation and image generation are distinct model output capabilities. The expected acceptance evidence is: The response includes the requested image artifact and metadata. This observation directly tests the stated behavior.

Incorrect Answers

B:

Language-model likelihood does not independently establish factual truth. This observation does not establish the required result: The response includes the requested image artifact and metadata.

C:

High accuracy can conceal failure to detect rare positive events. This observation does not establish the required result: The response includes the requested image artifact and metadata.

D:

Instruction roles help distinguish application policy from user task text. This observation does not establish the required result: The response includes the requested image artifact and metadata.

Question 739

A project needs the following AI behavior: An analyst needs a structured response from a video of a product demonstration. What should the designer implement?

  1. Use a video-capable Content Understanding analyzer with timestamps and supported fields
  2. Recompute arithmetic and compare totals with reported invoice fields
  3. Extract checkbox selection marks with supported document analysis, and verify the visible signature separately; use signature detection only if its preview API is approved
  4. Use layout-aware document analysis that preserves rows, selection marks and text spans

Explanation

Correct Answer

A:

Multimodal extraction draws on visual and spoken evidence over time. This implementation addresses the precise goal rather than an adjacent AI task.

Incorrect Answers

B:

Independent business rules catch plausible extraction errors. That is not sufficient here: Multimodal extraction draws on visual and spoken evidence over time.

C:

Selection-mark extraction is available with GA document analysis, but dedicated signature-location detection is available only in the 2026-06-01-preview API and cannot establish legal validity. That is not sufficient here: Multimodal extraction draws on visual and spoken evidence over time.

D:

Ordinary plain-text OCR may lose table relationships and checkbox states. That is not sufficient here: Multimodal extraction draws on visual and spoken evidence over time.

Question 740

Troubleshooting report: The assistant returns free-form prose that breaks an app parser. Which correction is MOST directly targeted to this failure?

  1. Require a completed tool result and error-path response
  2. Use the supported image-content input and multimodal deployment
  3. Tighten the output contract and apply post-response validation
  4. Keep policy instructions separate from lower-trust data

Explanation

Correct Answer

C:

This correction directly addresses the reported failure. Explicit examples can improve adherence to bounded output conventions. The targeted change is: Tighten the output contract and apply post-response validation.

Incorrect Answers

A:

Timeout handling should not be reported as a successful action. In the stated case, troubleshooting report: The assistant returns free-form prose that breaks an app parser; “Require a completed tool result and error-path response” does not target that failure.

B:

The model must receive the actual visual content to reason about it. In the stated case, troubleshooting report: The assistant returns free-form prose that breaks an app parser; “Use the supported image-content input and multimodal deployment” does not target that failure.

D:

High-priority instructions should describe approved behavior and escalation rules. In the stated case, troubleshooting report: The assistant returns free-form prose that breaks an app parser; “Keep policy instructions separate from lower-trust data” does not target that failure.

Microsoft Azure AI Fundamentals (AI-901) Practice Tests