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 38

Question 370

An Azure AI Fundamentals solution has this requirement: A Speech service client fails because its configured region differs from its resource. Which implementation most directly addresses it?

  1. Match the speech resource region or endpoint to the SDK configuration
  2. Evaluate word error rate on the same held-out audio samples
  3. Use word or utterance timestamps from a supported recognition output
  4. Combine recognition or audio understanding with verified text and text-to-speech

Explanation

Correct Answer

A:

Region or endpoint mismatch prevents requests reaching the correct resource. This implementation addresses the precise goal rather than an adjacent AI task.

Incorrect Answers

B:

Comparable labeled audio prevents misleading quality claims. That is not sufficient here: Region or endpoint mismatch prevents requests reaching the correct resource.

C:

Timecodes align text segments to audio playback. That is not sufficient here: Region or endpoint mismatch prevents requests reaching the correct resource.

D:

Input transcription and output synthesis are distinct steps. That is not sufficient here: Region or endpoint mismatch prevents requests reaching the correct resource.

Question 371

A solution team has selected this control: Use a generative language workload with policy-grounded response constraints. Which observation is the most direct acceptance evidence?

  1. The spoken message matches the approved written text
  2. The assistant produces a relevant answer with approved citations
  3. A prompt-only request yields a new original illustration rather than a classification label
  4. Customer cohorts have validated internal cohesion and actionable differences

Explanation

Correct Answer

B:

Generation produces novel text, unlike choosing a fixed class label. The expected acceptance evidence is: The assistant produces a relevant answer with approved citations. This observation directly tests the stated behavior.

Incorrect Answers

A:

Speech synthesis converts supplied text into audible output. This observation does not establish the required result: The assistant produces a relevant answer with approved citations.

C:

Text-to-image generation creates new visual content from a description; classification only assigns labels to existing inputs. This observation does not establish the required result: The assistant produces a relevant answer with approved citations.

D:

Clustering discovers groups based on feature similarity. This observation does not establish the required result: The assistant produces a relevant answer with approved citations.

Question 372

An AI operations team is testing two distinct capabilities. Failure (1): The team center-crops every frame and cuts off the subject. Failure (2): The app sends a generation request to an OCR API. Which TWO specific corrections should the team apply?

  1. Route to the text-to-image mode with intended inputs only
  2. Retain supported bounding polygons or field anchors
  3. Route it to an image generation deployment
  4. Apply confidence and nonmaximum suppression or instance deduplication
  5. Apply smart-crop suggestions or salient-region detection

Explanation

Correct Answers

C:

Correct for failure (2). Image generation creates new pixels rather than classifying an existing photo. It independently corrects the second defect.

E:

Correct for failure (1). Smart cropping attempts to retain salient regions within target dimensions. The proposed correction directly addresses that reported defect.

Incorrect Answers

A:

Generation from text is distinct from image editing. It does not complete the two decisions required in this question.

B:

Spatial provenance supports later verification of recognition results. It does not complete the two decisions required in this question.

D:

Object counts require resolving distinct instances, not trusting one scene label. The stated pair of failures instead requires “Route it to an image generation deployment” and “Apply smart-crop suggestions or salient-region detection”; this option cannot substitute for both.

Question 373

An acceptance tester needs to validate: Use text classification with labeled intent examples. Which observed result proves the intended behavior?

  1. The transcript contains the actual spoken issue and timestamp
  2. Held-out requests receive the appropriate routed intent labels
  3. A prompt-only request yields a new original illustration rather than a classification label
  4. Follow-up questions resolve consistently to the same approved case documents

Explanation

Correct Answer

B:

Classification maps text to one of several predefined categories. The expected acceptance evidence is: Held-out requests receive the appropriate routed intent labels. This observation directly tests the stated behavior.

Incorrect Answers

A:

Speech recognition converts speech into words suitable for search. This observation does not establish the required result: Held-out requests receive the appropriate routed intent labels.

C:

Text-to-image generation creates new visual content from a description; classification only assigns labels to existing inputs. This observation does not establish the required result: Held-out requests receive the appropriate routed intent labels.

D:

The assistant should incorporate context while remaining connected to reliable sources. This observation does not establish the required result: Held-out requests receive the appropriate routed intent labels.

Question 374

Troubleshooting report: The microphone buffer is cleared before recognition consumes it. Which correction is MOST directly targeted to this failure?

  1. Use the sentiment endpoint and validate its supported language
  2. Add an authorized synthesis stage for the answer
  3. Correct streaming buffer and session lifecycle
  4. Load the deployed speech resource's region and endpoint

Explanation

Correct Answer

C:

This correction directly addresses the reported failure. Real-time audio pipelines need coordinated capture and recognition start. The targeted change is: Correct streaming buffer and session lifecycle.

Incorrect Answers

A:

Sentiment output classifies attitudes expressed in the submitted text. In the stated case, troubleshooting report: The microphone buffer is cleared before recognition consumes it; “Use the sentiment endpoint and validate its supported language” does not target that failure.

B:

Input transcription and output synthesis are distinct steps. In the stated case, troubleshooting report: The microphone buffer is cleared before recognition consumes it; “Add an authorized synthesis stage for the answer” does not target that failure.

D:

Region or endpoint mismatch prevents requests reaching the correct resource. In the stated case, troubleshooting report: The microphone buffer is cleared before recognition consumes it; “Load the deployed speech resource's region and endpoint” does not target that failure.

Question 375

An engineering review documents this requirement: An AI app repeatedly truncates an otherwise correct long response. Select the most appropriate solution.

  1. Require verifiable grounding and calibrated uncertainty
  2. Use structured outputs or strict application-side schema validation
  3. Benchmark both on representative quality, latency and cost thresholds
  4. Check output-token limits and context-budget allocation

Explanation

Correct Answer

D:

Generation stops when an output budget or stop condition is reached. This implementation addresses the precise goal rather than an adjacent AI task.

Incorrect Answers

A:

Language-model likelihood does not independently establish factual truth. That is not sufficient here: Generation stops when an output budget or stop condition is reached.

B:

Natural-language fluency is not a guarantee of machine-readable contract validity. That is not sufficient here: Generation stops when an output budget or stop condition is reached.

C:

Model size alone does not establish suitability for a particular task. That is not sufficient here: Generation stops when an output budget or stop condition is reached.

Question 376

The approved design action is: Use a supported dense-captioning capability where available. Which test result would BEST establish that it worked?

  1. Prominent objects receive separate region captions and coordinates
  2. A labeled photograph receives an accurate high-level summary
  3. Expected objects have correctly positioned boxes
  4. The result includes supported text, tags and person boxes without conflation

Explanation

Correct Answer

A:

Dense captions describe localized regions beyond a single global caption. The expected acceptance evidence is: Prominent objects receive separate region captions and coordinates. This observation directly tests the stated behavior.

Incorrect Answers

B:

Captioning produces a sentence-level description rather than a coordinate list. This observation does not establish the required result: Prominent objects receive separate region captions and coordinates.

C:

Object detection reports which objects appear and where they occur. This observation does not establish the required result: Prominent objects receive separate region captions and coordinates.

D:

Different visual outputs describe text, scene classes and detected people. This observation does not establish the required result: Prominent objects receive separate region captions and coordinates.

Question 377

Troubleshooting report: The system produces approximate figures when the dataset is unavailable. Which correction is MOST directly targeted to this failure?

  1. Partition state by user and clear or secure session context
  2. Filter sources by effective status before similarity ranking
  3. Audit correlated features and document the business necessity of each factor
  4. Require a data query or report unavailability

Explanation

Correct Answer

D:

This correction directly addresses the reported failure. A numeric-looking answer is not proof that a measure was computed. The targeted change is: Require a data query or report unavailability.

Incorrect Answers

A:

Privacy includes preventing cross-session memory disclosure. In the stated case, troubleshooting report: The system produces approximate figures when the dataset is unavailable; “Partition state by user and clear or secure session context” does not target that failure.

B:

Truthfulness requires evidence that corresponds to the operative version. In the stated case, troubleshooting report: The system produces approximate figures when the dataset is unavailable; “Filter sources by effective status before similarity ranking” does not target that failure.

C:

Legitimate credit criteria cannot be replaced by sensitive identity traits. In the stated case, troubleshooting report: The system produces approximate figures when the dataset is unavailable; “Audit correlated features and document the business necessity of each factor” does not target that failure.

Question 378

Troubleshooting report: The system cites a fabricated document title. Which correction is MOST directly targeted to this failure?

  1. Keep temporal and content provenance in the retrieval result
  2. Build citations from stored IDs and source spans
  3. Verify the actual source signature region or use explicitly approved preview detection, never infer signing from a typed name
  4. Configure required invoice fields and validate extracted values

Explanation

Correct Answer

B:

This correction directly addresses the reported failure. Verifiable provenance differentiates a cited answer from guesswork. The targeted change is: Build citations from stored IDs and source spans.

Incorrect Answers

A:

Multimodal grounding should preserve how the answer maps to source media. In the stated case, troubleshooting report: The system cites a fabricated document title; “Keep temporal and content provenance in the retrieval result” does not target that failure.

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. In the stated case, troubleshooting report: The system cites a fabricated document title; “Verify the actual source signature region or use explicitly approved preview detection, never infer signing from a typed name” does not target that failure.

D:

Document extraction turns unstructured pages into validated structured fields. In the stated case, troubleshooting report: The system cites a fabricated document title; “Configure required invoice fields and validate extracted values” does not target that failure.

Question 379

A project needs the following AI behavior: An AI deployment has no named owner for harmful production behavior. What should the designer implement?

  1. Identify the virtual agent as synthetic and avoid deceptive identity cues
  2. Assign accountable operational and risk owners with escalation responsibilities
  3. Version procedures and limit grounding to the latest authorized operational guidance
  4. Test representative document layouts and accessibility-oriented inputs

Explanation

Correct Answer

B:

An AI model cannot be its own accountable incident responder. This implementation addresses the precise goal rather than an adjacent AI task.

Incorrect Answers

A:

Transparency prevents users from falsely assuming they are speaking with a human. That is not sufficient here: An AI model cannot be its own accountable incident responder.

C:

A precise explanation of an obsolete procedure can still be dangerous. That is not sufficient here: An AI model cannot be its own accountable incident responder.

D:

Inclusion requires robust input handling beyond a single document template. That is not sufficient here: An AI model cannot be its own accountable incident responder.

Question 380

An Azure AI solution uses the following approved control: Detect contradictory extracted values and route uncertain cases for review. What observable result should pass QA?

  1. The answer resolves to a specific approved source document version
  2. A reply links its claim to a transcript time or relevant video segment
  3. The result reports both candidate values and cites their locations
  4. The output returns correct limits with supporting clauses

Explanation

Correct Answer

C:

A model should not silently choose among competing contractual evidence. The expected acceptance evidence is: The result reports both candidate values and cites their locations. This observation directly tests the stated behavior.

Incorrect Answers

A:

Verifiable provenance differentiates a cited answer from guesswork. This observation does not establish the required result: The result reports both candidate values and cites their locations.

B:

Multimodal grounding should preserve how the answer maps to source media. This observation does not establish the required result: The result reports both candidate values and cites their locations.

D:

Custom extraction maps semantically relevant phrases to typed fields. This observation does not establish the required result: The result reports both candidate values and cites their locations.

Microsoft Azure AI Fundamentals (AI-901) Practice Tests