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 34

Question 330

Troubleshooting report: The app assumes one image caption contains all structured outputs. Which correction is MOST directly targeted to this failure?

  1. Request the needed visual features explicitly
  2. Route to the text-to-image mode with intended inputs only
  3. Add confidence gating and supporting evidence
  4. Use aspect-ratio-preserving crop or padding

Explanation

Correct Answer

A:

This correction directly addresses the reported failure. Different visual outputs describe text, scene classes and detected people. The targeted change is: Request the needed visual features explicitly.

Incorrect Answers

B:

Generation from text is distinct from image editing. In the stated case, troubleshooting report: The app assumes one image caption contains all structured outputs; “Route to the text-to-image mode with intended inputs only” does not target that failure.

C:

Generated visual interpretations can be wrong even when phrased fluently. In the stated case, troubleshooting report: The app assumes one image caption contains all structured outputs; “Add confidence gating and supporting evidence” does not target that failure.

D:

Smart cropping differs from simply stretching pixels. In the stated case, troubleshooting report: The app assumes one image caption contains all structured outputs; “Use aspect-ratio-preserving crop or padding” does not target that failure.

Question 331

An Azure AI Fundamentals solution has this requirement: An engineering group needs a novel illustration from a detailed product description. Which implementation most directly addresses it?

  1. Use a relevance-ranking or recommendation model evaluated on resolved-case history
  2. Use regression or forecasting appropriate to time-series data
  3. Identify a text-to-image generative workload, not an existing-image classifier
  4. Use PII-aware entity recognition with appropriate redaction

Explanation

Correct Answer

C:

Text-to-image generation creates new visual content from a description; classification only assigns labels to existing inputs.

Incorrect Answers

A:

Recommendations rank candidate items rather than simply generating arbitrary titles. That is not sufficient here: Text-to-image generation creates new visual content from a description; classification only assigns labels to existing inputs.

B:

Forecasts predict future numeric quantities rather than discrete tags. That is not sufficient here: Text-to-image generation creates new visual content from a description; classification only assigns labels to existing inputs.

D:

Entity extraction locates sensitive spans for privacy handling. That is not sufficient here: Text-to-image generation creates new visual content from a description; classification only assigns labels to existing inputs.

Question 332

An AI operations team is testing two distinct capabilities. Failure (1): The application assumes every invoice has totals at one fixed pixel coordinate. Failure (2): The system cites a fabricated document title. Which TWO specific corrections should the team apply?

  1. Enable appropriate diarization and handle ambiguous overlap
  2. Use layout-aware field definitions rather than fixed crop positions
  3. Build citations from stored IDs and source spans
  4. Use the supported GA analyzer configuration
  5. Retain page IDs and reconcile source order

Explanation

Correct Answers

B:

Correct for failure (1). A schema-driven analyzer identifies business fields independent of fixed coordinates. The proposed correction directly addresses that reported defect.

C:

Correct for failure (2). Verifiable provenance differentiates a cited answer from guesswork. It independently corrects the second defect.

Incorrect Answers

A:

Conversation extraction needs more than a single flattened text stream. The stated pair of failures instead requires “Use layout-aware field definitions rather than fixed crop positions” and “Build citations from stored IDs and source spans”; this option cannot substitute for both.

D:

Preview-only features may have different availability and contracts. It does not complete the two decisions required in this question.

E:

Page order and references are part of document provenance. The stated pair of failures instead requires “Use layout-aware field definitions rather than fixed crop positions” and “Build citations from stored IDs and source spans”; this option cannot substitute for both.

Question 333

An engineering review documents this requirement: A meeting assistant needs concise action items from a recognized transcript. Select the most appropriate solution.

  1. Apply named entity recognition and preserve detected spans and categories
  2. Stop active recording and apply the documented microphone permission state
  3. Convert the input to a format and sample rate supported by the chosen API
  4. Run grounded summarization after speech recognition and preserve source segments

Explanation

Correct Answer

D:

Transcription obtains words; summarization extracts decisions and tasks. This implementation addresses the precise goal rather than an adjacent AI task.

Incorrect Answers

A:

NER identifies named spans rather than summarizing the email. That is not sufficient here: Transcription obtains words; summarization extracts decisions and tasks.

B:

User control of capture must be enforced at the device or application layer. That is not sufficient here: Transcription obtains words; summarization extracts decisions and tasks.

C:

Speech recognition requires the service to decode audio correctly. That is not sufficient here: Transcription obtains words; summarization extracts decisions and tasks.

Question 334

A solution team has selected this control: Enable a supported barcode extraction or recognition feature. Which observation is the most direct acceptance evidence?

  1. The lot code matches the visible label and its image region
  2. The missing field is returned as null or flagged unknown
  3. Several labeled forms populate the same typed output fields
  4. The decoded barcode matches the printed test label

Explanation

Correct Answer

D:

A barcode is a structured visual symbol not ordinary OCR text. The expected acceptance evidence is: The decoded barcode matches the printed test label. This observation directly tests the stated behavior.

Incorrect Answers

A:

Visual information extraction can map pixels to structured identifiers. This observation does not establish the required result: The decoded barcode matches the printed test label.

B:

Extraction should preserve absence and distinguish it from zero. This observation does not establish the required result: The decoded barcode matches the printed test label.

C:

A schema-driven analyzer identifies business fields independent of fixed coordinates. This observation does not establish the required result: The decoded barcode matches the printed test label.

Question 335

The approved design action is: Record the actual decision output, input reference, model and prompt versions in integrity-protected audit records. Which test result would BEST establish that it worked?

  1. A simulated timeout yields a failure status without a fabricated numeric balance
  2. Scanner-specific evaluation identifies the weak cohort and blocks unsupported rollout
  3. An auditor retrieves the archived recommendation and its corresponding versioned input and configuration
  4. Missing dosage context triggers abstention and a review route

Explanation

Correct Answer

C:

Auditability requires the original decision evidence and version identifiers; regenerating model text is not guaranteed to be identical. The expected acceptance evidence is: An auditor retrieves the archived recommendation and its corresponding versioned input and configuration. This observation directly tests the stated behavior.

Incorrect Answers

A:

A fluent generation cannot substitute for missing authoritative transaction data. This observation does not establish the required result: An auditor retrieves the archived recommendation and its corresponding versioned input and configuration.

B:

Reliability requires detecting distribution shifts hidden in aggregate accuracy. This observation does not establish the required result: An auditor retrieves the archived recommendation and its corresponding versioned input and configuration.

D:

Safety-critical recommendations require verified constraints outside open-ended generation. This observation does not establish the required result: An auditor retrieves the archived recommendation and its corresponding versioned input and configuration.

Question 336

An AI operations team is testing two distinct capabilities. Failure (1): The app sends the waveform to a text-only inference method. Failure (2): The team calls speech-to-text on already written text. Which TWO specific corrections should the team apply?

  1. Use a speech synthesis endpoint and compatible output format
  2. Review label provenance and downstream decision disparities
  3. Refresh indexed grounding sources and verify their effective date
  4. Pin compatible deployment settings and maintain versioned evaluations
  5. Choose a recognized audio-input interface and accepted encoding

Explanation

Correct Answers

A:

Correct for failure (2). Speech synthesis generates audio from text rather than transcribing audio. It independently corrects the second defect.

E:

Correct for failure (1). Model capabilities must include the required audio modality before the client can obtain speech-to-text results. The proposed correction directly addresses that reported defect.

Incorrect Answers

B:

Bias can enter through the targets themselves, not just inputs. The stated pair of failures instead requires “Use a speech synthesis endpoint and compatible output format” and “Choose a recognized audio-input interface and accepted encoding”; this option cannot substitute for both.

C:

External retrieval updates factual context without retraining model weights. The stated pair of failures instead requires “Use a speech synthesis endpoint and compatible output format” and “Choose a recognized audio-input interface and accepted encoding”; this option cannot substitute for both.

D:

Reproducibility depends on more than the visible prompt text. The stated pair of failures instead requires “Use a speech synthesis endpoint and compatible output format” and “Choose a recognized audio-input interface and accepted encoding”; this option cannot substitute for both.

Question 337

The approved design action is: Benchmark both on representative quality, latency and cost thresholds. Which test result would BEST establish that it worked?

  1. The chosen passages fit the model budget while preserving needed evidence
  2. The model answers a question using the actual supplied photo and text
  3. The smaller candidate meets the acceptance floor with lower cost where appropriate
  4. The holdout report identifies missed positives and false alarms explicitly

Explanation

Correct Answer

C:

Model size alone does not establish suitability for a particular task. The expected acceptance evidence is: The smaller candidate meets the acceptance floor with lower cost where appropriate. This observation directly tests the stated behavior.

Incorrect Answers

A:

A finite context window limits how much material can be processed at once. This observation does not establish the required result: The smaller candidate meets the acceptance floor with lower cost where appropriate.

B:

Multimodal models can reason over more than one supported input modality. This observation does not establish the required result: The smaller candidate meets the acceptance floor with lower cost where appropriate.

D:

High accuracy can conceal failure to detect rare positive events. This observation does not establish the required result: The smaller candidate meets the acceptance floor with lower cost where appropriate.

Question 338

Troubleshooting report: An opaque derived score lacks documented inputs. Which correction is MOST directly targeted to this failure?

  1. Apply data minimization, retention and permission controls
  2. Propagate typed tool errors and use an explicit unavailable response
  3. Add subgroup and matched-pair fairness tests before approval
  4. Document feature lineage and remove unapproved decision features

Explanation

Correct Answer

D:

This correction directly addresses the reported failure. Accountability requires interpretable reasons for consequential decisions. The targeted change is: Document feature lineage and remove unapproved decision features.

Incorrect Answers

A:

Privacy requires meaningful notice and bounded handling of personal data. In the stated case, troubleshooting report: An opaque derived score lacks documented inputs; “Apply data minimization, retention and permission controls” does not target that failure.

B:

A fluent generation cannot substitute for missing authoritative transaction data. In the stated case, troubleshooting report: An opaque derived score lacks documented inputs; “Propagate typed tool errors and use an explicit unavailable response” does not target that failure.

C:

Fairness checks compare materially equivalent cases rather than raw group averages alone. In the stated case, troubleshooting report: An opaque derived score lacks documented inputs; “Add subgroup and matched-pair fairness tests before approval” does not target that failure.

Question 339

A solution team has selected this control: Expose a read-only tool contract rather than a broad administrative API. Which observation is the most direct acceptance evidence?

  1. The test output satisfies every defined requirement of the prompt
  2. The same prompt revision is applied across test deployments
  3. A list operation succeeds and a mutation attempt is denied
  4. An injected page fails to invoke an unauthorized tool

Explanation

Correct Answer

C:

Least-privilege tool methods reduce accidental external actions. The expected acceptance evidence is: A list operation succeeds and a mutation attempt is denied. This observation directly tests the stated behavior.

Incorrect Answers

A:

Specific instructions reduce ambiguity about what constitutes success. This observation does not establish the required result: A list operation succeeds and a mutation attempt is denied.

B:

Versioned system messages provide a controlled behavioral baseline. This observation does not establish the required result: A list operation succeeds and a mutation attempt is denied.

D:

External content cannot replace privileged system behavior. This observation does not establish the required result: A list operation succeeds and a mutation attempt is denied.

Question 340

An engineering review documents this requirement: A fleet operator wants to predict whether a vehicle will fail in the next week. Select the most appropriate solution.

  1. Use retrieval-augmented generation with actual source citations
  2. Use a visual classification or detection model appropriate to whether location matters
  3. Use supervised classification with labeled maintenance outcomes
  4. Use sentiment analysis to estimate positive, negative, or neutral tone

Explanation

Correct Answer

C:

A failure probability or class is a discrete decision under a defined horizon. This implementation addresses the precise goal rather than an adjacent AI task.

Incorrect Answers

A:

RAG retrieves current external evidence for a generated response. That is not sufficient here: A failure probability or class is a discrete decision under a defined horizon.

B:

The task chooses a label, with localization needed only when multiple signs appear. That is not sufficient here: A failure probability or class is a discrete decision under a defined horizon.

D:

Sentiment analysis infers expressed attitude rather than extracting names. That is not sufficient here: A failure probability or class is a discrete decision under a defined horizon.

Microsoft Azure AI Fundamentals (AI-901) Practice Tests