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 8

Question 70

A project needs the following AI behavior: A planned peak workload exceeds the selected model deployment throughput quota before launch. What should the designer implement?

  1. Choose a model or pipeline that supports both visual and audio modalities
  2. Choose a speech-recognition model or audio-capable multimodal endpoint with a documented audio-input contract
  3. Measure expected tokens and requests per minute, then choose an approved throughput tier or capacity adjustment before launch
  4. Choose a compatible multimodal model and provide supported text/image inputs

Explanation

Correct Answer

C:

Quota sizing is a deployment-capacity decision; retrying a predictable sustained excess cannot add model throughput.

Incorrect Answers

A:

Different media channels provide complementary scene and dialogue evidence. That is not sufficient here: Quota sizing is a deployment-capacity decision; retrying a predictable sustained excess cannot add model throughput.

B:

Model capabilities must include the required audio modality before the client can obtain speech-to-text results. That is not sufficient here: Quota sizing is a deployment-capacity decision; retrying a predictable sustained excess cannot add model throughput.

D:

Multimodal models can reason over more than one supported input modality. That is not sufficient here: Quota sizing is a deployment-capacity decision; retrying a predictable sustained excess cannot add model throughput.

Question 71

An Azure AI Fundamentals solution has this requirement: A video document summarizes an interview and must preserve the speaker's exact quoted words. Which implementation most directly addresses it?

  1. Use document extraction with source positions before indexing for search
  2. Preserve page identifiers and validate page sequence during extraction
  3. Distinguish transcript passages from generated summaries and retain timestamps
  4. Preserve document IDs, versions and extracted field evidence with search results

Explanation

Correct Answer

C:

Generated summaries are paraphrases and not guaranteed verbatim quotes. This implementation addresses the precise goal rather than an adjacent AI task.

Incorrect Answers

A:

Structured layout and provenance let users verify a quoted clause. That is not sufficient here: Generated summaries are paraphrases and not guaranteed verbatim quotes.

B:

Page order and references are part of document provenance. That is not sufficient here: Generated summaries are paraphrases and not guaranteed verbatim quotes.

D:

Verifiable provenance differentiates a cited answer from guesswork. That is not sufficient here: Generated summaries are paraphrases and not guaranteed verbatim quotes.

Question 72

Troubleshooting report: The system simply tags the sentence as positive. Which correction is MOST directly targeted to this failure?

  1. Use entity recognition and verify sensitive-entity recall
  2. Invoke image segmentation and compositing or a controlled image editor
  3. Call the authorized ticket tool and verify the receipt
  4. Ground the summary in extracted transcript segments

Explanation

Correct Answer

A:

This correction directly addresses the reported failure. Entity extraction locates sensitive spans for privacy handling. The targeted change is: Use entity recognition and verify sensitive-entity recall.

Incorrect Answers

B:

Editing changes selected pixels while leaving product regions intact. In the stated case, troubleshooting report: The system simply tags the sentence as positive; “Invoke image segmentation and compositing or a controlled image editor” does not target that failure.

C:

Agentic workloads combine reasoning and bounded external actions. In the stated case, troubleshooting report: The system simply tags the sentence as positive; “Call the authorized ticket tool and verify the receipt” does not target that failure.

D:

Meeting summaries require temporal and spoken source content. In the stated case, troubleshooting report: The system simply tags the sentence as positive; “Ground the summary in extracted transcript segments” does not target that failure.

Question 73

The approved design action is: Provide tested style instructions and examples in the prompt before considering tuning. Which test result would BEST establish that it worked?

  1. Repeated structured cases follow the expected schema more consistently
  2. The same facts are consistently presented in the required style
  3. Semantically related sentences have nearer vectors in an evaluated sample
  4. The summary references events visible on frames and spoken in the transcript

Explanation

Correct Answer

B:

Clear prompting can control tone without model retraining. The expected acceptance evidence is: The same facts are consistently presented in the required style. This observation directly tests the stated behavior.

Incorrect Answers

A:

Lower randomness often improves consistency but does not guarantee correctness. This observation does not establish the required result: The same facts are consistently presented in the required style.

C:

Embeddings encode learned semantic relationships rather than literal string identity. This observation does not establish the required result: The same facts are consistently presented in the required style.

D:

Different media channels provide complementary scene and dialogue evidence. This observation does not establish the required result: The same facts are consistently presented in the required style.

Question 74

An Azure AI Fundamentals solution has this requirement: An autonomous procurement tool has broader API permissions than its task needs. Which implementation most directly addresses it?

  1. Provide semantic labels and keyboard-operable controls for all important interactions
  2. Evaluate safety again after tool and retrieval results are incorporated
  3. Record decision reasons and give a human review route for important outcomes
  4. Restrict service credentials to a tested allowlist of permitted operations

Explanation

Correct Answer

D:

Least privilege reduces damage from compromised prompts or tool output. This implementation addresses the precise goal rather than an adjacent AI task.

Incorrect Answers

A:

Inclusive design permits nonvisual navigation of the same task path. That is not sufficient here: Least privilege reduces damage from compromised prompts or tool output.

B:

Threats can enter later stages even when the original user request was harmless. That is not sufficient here: Least privilege reduces damage from compromised prompts or tool output.

C:

Accountability includes traceable responsibility and appropriate contestability. That is not sufficient here: Least privilege reduces damage from compromised prompts or tool output.

Question 75

Troubleshooting report: The app returns only a generic paragraph about the invoice. Which correction is MOST directly targeted to this failure?

  1. Inspect multiple temporal segments and independently review brief events that sampled frames could miss
  2. Represent repeated fields as arrays and validate counts
  3. Configure required invoice fields and validate extracted values
  4. Verify the actual source signature region or use explicitly approved preview detection, never infer signing from a typed name

Explanation

Correct Answer

C:

This correction directly addresses the reported failure. Document extraction turns unstructured pages into validated structured fields. The targeted change is: Configure required invoice fields and validate extracted values.

Incorrect Answers

A:

Content Understanding samples approximately one frame per second, so brief or single-frame actions can be missed despite valid timestamps. In the stated case, troubleshooting report: The app returns only a generic paragraph about the invoice; “Inspect multiple temporal segments and independently review brief events that sampled frames could miss” does not target that failure.

B:

Repeated records require structured lists with item-level associations. In the stated case, troubleshooting report: The app returns only a generic paragraph about the invoice; “Represent repeated fields as arrays and validate counts” does not target that failure.

D:

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 app returns only a generic paragraph about the invoice; “Verify the actual source signature region or use explicitly approved preview detection, never infer signing from a typed name” does not target that failure.

Question 76

Troubleshooting report: Raw nonimage bytes are sent with an image MIME label. Which correction is MOST directly targeted to this failure?

  1. Collect balanced labeled examples and test class confusion
  2. Route to the text-to-image mode with intended inputs only
  3. Use aspect-ratio-preserving crop or padding
  4. Decode and re-encode to a supported image format

Explanation

Correct Answer

D:

This correction directly addresses the reported failure. Vision endpoints impose supported input formats and limits. The targeted change is: Decode and re-encode to a supported image format.

Incorrect Answers

A:

Fine-grained visual classes require representative discriminating images. In the stated case, troubleshooting report: Raw nonimage bytes are sent with an image MIME label; “Collect balanced labeled examples and test class confusion” does not target that failure.

B:

Generation from text is distinct from image editing. In the stated case, troubleshooting report: Raw nonimage bytes are sent with an image MIME label; “Route to the text-to-image mode with intended inputs only” does not target that failure.

C:

Smart cropping differs from simply stretching pixels. In the stated case, troubleshooting report: Raw nonimage bytes are sent with an image MIME label; “Use aspect-ratio-preserving crop or padding” does not target that failure.

Question 77

A project needs the following AI behavior: A medical dictation app must recognize specialty vocabulary. What should the designer implement?

  1. Use speech-to-text recognition on a supported audio input stream
  2. Check playback device selection and the synthesized audio format
  3. Use a supported diarization or speaker-labeling capability where available
  4. Evaluate custom phrase hints or supported speech customization on representative terms

Explanation

Correct Answer

D:

Domain-specific words may require biasing or adaptation beyond a generic recognizer. This implementation addresses the precise goal rather than an adjacent AI task.

Incorrect Answers

A:

Speech recognition converts waveform features to text tokens. That is not sufficient here: Domain-specific words may require biasing or adaptation beyond a generic recognizer.

B:

Successful synthesis does not guarantee local audio is routed to a speaker. That is not sufficient here: Domain-specific words may require biasing or adaptation beyond a generic recognizer.

C:

Speaker diarization separates turns and identities at a conversation level. That is not sufficient here: Domain-specific words may require biasing or adaptation beyond a generic recognizer.

Question 78

An acceptance tester needs to validate: Use regression or forecasting appropriate to time-series data. Which observed result proves the intended behavior?

  1. The result contains product boxes in their correct shelf positions
  2. The timeline marks arrival frames with appropriate timestamps
  3. Extracted checks and notes match visible form locations
  4. Predicted consumption is evaluated against observed later readings

Explanation

Correct Answer

D:

Forecasts predict future numeric quantities rather than discrete tags. The expected acceptance evidence is: Predicted consumption is evaluated against observed later readings. This observation directly tests the stated behavior.

Incorrect Answers

A:

Object detectors return classes and spatial boundaries. This observation does not establish the required result: Predicted consumption is evaluated against observed later readings.

B:

Video understanding retains order and time, unlike single-image tagging. This observation does not establish the required result: Predicted consumption is evaluated against observed later readings.

C:

Different form elements require more than ordinary running-text OCR. This observation does not establish the required result: Predicted consumption is evaluated against observed later readings.

Question 79

An acceptance tester needs to validate: Choose a text-classification or sentiment-analysis capability rather than an image generator. Which observed result proves the intended behavior?

  1. The returned text includes the photographed serial number
  2. The output returns evaluated sentiment classes rather than generated images
  3. The same facts are consistently presented in the required style
  4. The response includes the requested image artifact and metadata

Explanation

Correct Answer

B:

Different model tasks require compatible outputs and labeled evaluation data. The expected acceptance evidence is: The output returns evaluated sentiment classes rather than generated images. This observation directly tests the stated behavior.

Incorrect Answers

A:

OCR produces readable text from pixels. This observation does not establish the required result: The output returns evaluated sentiment classes rather than generated images.

C:

Clear prompting can control tone without model retraining. This observation does not establish the required result: The output returns evaluated sentiment classes rather than generated images.

D:

Text generation and image generation are distinct model output capabilities. This observation does not establish the required result: The output returns evaluated sentiment classes rather than generated images.

Question 80

Troubleshooting report: The agent treats the user text as a new system instruction. Which correction is MOST directly targeted to this failure?

  1. Preserve role separation and tool-side permission checks
  2. Add idempotency keys and query the original transaction state
  3. Version prompt and deployment settings and retain the actual evaluation output
  4. Rotate the key and load credentials through a governed secret mechanism

Explanation

Correct Answer

A:

This correction directly addresses the reported failure. A user cannot use text to elevate permissions beyond the application policy. The targeted change is: Preserve role separation and tool-side permission checks.

Incorrect Answers

B:

A timeout does not prove that the original mutation failed. In the stated case, troubleshooting report: The agent treats the user text as a new system instruction; “Add idempotency keys and query the original transaction state” does not target that failure.

C:

Recorded settings support investigation, but generative models can still vary on an apparently identical rerun. In the stated case, troubleshooting report: The agent treats the user text as a new system instruction; “Version prompt and deployment settings and retain the actual evaluation output” does not target that failure.

D:

Credentials should not appear in repositories, prompt logs or client binaries. In the stated case, troubleshooting report: The agent treats the user text as a new system instruction; “Rotate the key and load credentials through a governed secret mechanism” does not target that failure.

Microsoft Azure AI Fundamentals (AI-901) Practice Tests