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 47

Question 460

An acceptance tester needs to validate: Use operation IDs and read-back receipts before replaying external side effects. Which observed result proves the intended behavior?

  1. A trace shows which tool result supported each claimed business action
  2. Execution traces show the intended tool invoked with valid arguments
  3. Repeated events yield at most one order creation under the same operation ID
  4. A repository scan reveals no hardcoded secret and the app still authenticates

Explanation

Correct Answer

C:

A timeout does not prove that the original mutation failed. The expected acceptance evidence is: Repeated events yield at most one order creation under the same operation ID. This observation directly tests the stated behavior.

Incorrect Answers

A:

Operational observability aids debugging without exposing hidden sensitive prompt data. This observation does not establish the required result: Repeated events yield at most one order creation under the same operation ID.

B:

Tools must be discoverable with valid schemas before selection. This observation does not establish the required result: Repeated events yield at most one order creation under the same operation ID.

D:

Credentials should not appear in repositories, prompt logs or client binaries. This observation does not establish the required result: Repeated events yield at most one order creation under the same operation ID.

Question 461

Troubleshooting report: The preview has distorted people after unconstrained stretching. Which correction is MOST directly targeted to this failure?

  1. Ground the alt text in the actual picture
  2. Use the documented feature list for the deployed API version
  3. Route it to an image generation deployment
  4. Use aspect-ratio-preserving crop or padding

Explanation

Correct Answer

D:

This correction directly addresses the reported failure. Smart cropping differs from simply stretching pixels. The targeted change is: Use aspect-ratio-preserving crop or padding.

Incorrect Answers

A:

Accessible captions should communicate image purpose without invented details. In the stated case, troubleshooting report: The preview has distorted people after unconstrained stretching; “Ground the alt text in the actual picture” does not target that failure.

B:

Feature names and availability vary across versions. In the stated case, troubleshooting report: The preview has distorted people after unconstrained stretching; “Use the documented feature list for the deployed API version” does not target that failure.

C:

Image generation creates new pixels rather than classifying an existing photo. In the stated case, troubleshooting report: The preview has distorted people after unconstrained stretching; “Route it to an image generation deployment” does not target that failure.

Question 462

An AI operations team is testing two distinct capabilities. Failure (1): The app executes a safety shutdown solely from a weak visual guess. Failure (2): A global caption describes only the image background. Which TWO specific corrections should the team apply?

  1. Move sensitive images to authenticated private storage
  2. Crop or request region-focused visual analysis
  3. Add confidence gating and supporting evidence
  4. Review multiple timestamped frames and flag rapid or unobserved steps instead of inferring them
  5. Request tags and parse the tag collection

Explanation

Correct Answers

B:

Correct for failure (2). Specific regions can carry details missed by a whole-image caption. It independently corrects the second defect.

C:

Correct for failure (1). Generated visual interpretations can be wrong even when phrased fluently. The proposed correction directly addresses that reported defect.

Incorrect Answers

A:

Images can carry personally identifying information beyond text fields. The stated pair of failures instead requires “Crop or request region-focused visual analysis” and “Add confidence gating and supporting evidence”; this option cannot substitute for both.

D:

A single still cannot convey action order; sampled video frames may also omit fleeting intermediate steps. The stated pair of failures instead requires “Crop or request region-focused visual analysis” and “Add confidence gating and supporting evidence”; this option cannot substitute for both.

E:

This action would address another problem: the service returns paragraphs instead of requested tag labels. Tags summarize recognized concepts as labels. It does not resolve either specified failure.

Question 463

An Azure AI solution uses the following approved control: Use layout-aware document analysis that preserves rows, selection marks and text spans. What observable result should pass QA?

  1. Extracted rows align with the original form cells
  2. The checked option is returned accurately as true or selected
  3. Each product is returned as a separate item with its quantity
  4. Each date field links to the correct labeled region

Explanation

Correct Answer

A:

Ordinary plain-text OCR may lose table relationships and checkbox states. The expected acceptance evidence is: Extracted rows align with the original form cells. This observation directly tests the stated behavior.

Incorrect Answers

B:

Checkbox state is a structured signal rather than arbitrary prose. This observation does not establish the required result: Extracted rows align with the original form cells.

C:

Repeated records require structured lists with item-level associations. This observation does not establish the required result: Extracted rows align with the original form cells.

D:

Typed extraction avoids mixing semantically different dates. This observation does not establish the required result: Extracted rows align with the original form cells.

Question 464

An Azure AI Fundamentals solution has this requirement: An appointment chatbot serves patients using different languages. Which implementation most directly addresses it?

  1. Require governance evidence with reviewers, test outcomes, and documented exceptions
  2. Evaluate instructions and fallback quality in each supported language
  3. Verify permissions through the identity provider and server-side roles
  4. Use grounded medical content, high-risk detection and appropriate expert escalation

Explanation

Correct Answer

B:

Inclusion requires equivalent safety and meaning across supported locales. This implementation addresses the precise goal rather than an adjacent AI task.

Incorrect Answers

A:

Self-attestation is not independent proof of regulatory controls. That is not sufficient here: Inclusion requires equivalent safety and meaning across supported locales.

C:

Natural-language claims cannot elevate authentication authority. That is not sufficient here: Inclusion requires equivalent safety and meaning across supported locales.

D:

Safety mechanisms should reduce unsupported high-impact recommendations. That is not sufficient here: Inclusion requires equivalent safety and meaning across supported locales.

Question 465

An Azure AI Fundamentals solution has this requirement: A developer must choose between a compact model and a larger capable model. Which implementation most directly addresses it?

  1. Benchmark both on representative quality, latency and cost thresholds
  2. Use text-to-speech synthesis with a supported voice
  3. Represent text with embeddings and compare vectors using an appropriate similarity measure
  4. Explain token-based next-token prediction conditioned on preceding context

Explanation

Correct Answer

A:

Model size alone does not establish suitability for a particular task. This implementation addresses the precise goal rather than an adjacent AI task.

Incorrect Answers

B:

Speech synthesis generates audio from text rather than transcribing audio. That is not sufficient here: Model size alone does not establish suitability for a particular task.

C:

Embeddings encode learned semantic relationships rather than literal string identity. That is not sufficient here: Model size alone does not establish suitability for a particular task.

D:

A generative language model estimates likely continuations rather than reading a stored answer paragraph. That is not sufficient here: Model size alone does not establish suitability for a particular task.

Question 466

An acceptance tester needs to validate: Use text-to-speech synthesis with a supported voice. Which observed result proves the intended behavior?

  1. A peak-load forecast matches the configured token/request limits or identifies an approved capacity increase
  2. The output is produced as a sequence of tokens conditioned on the prompt
  3. A typed sentence is rendered as understandable speech
  4. A held-out set receives the correct whole-image classes and empirically calibrated class probabilities

Explanation

Correct Answer

C:

Speech synthesis generates audio from text rather than transcribing audio. The expected acceptance evidence is: A typed sentence is rendered as understandable speech. This observation directly tests the stated behavior.

Incorrect Answers

A:

Quota sizing is a deployment-capacity decision; retrying a predictable sustained excess cannot add model throughput. This observation does not establish the required result: A typed sentence is rendered as understandable speech.

B:

A generative language model estimates likely continuations rather than reading a stored answer paragraph. This observation does not establish the required result: A typed sentence is rendered as understandable speech.

D:

Classification assigns image-wide categories rather than bounding-box coordinates. This observation does not establish the required result: A typed sentence is rendered as understandable speech.

Question 467

An AI operations team is testing two distinct capabilities. Failure (1): The team fabricates purchase labels to run a supervised classifier. Failure (2): The same examples are used for training and claimed as unbiased testing. Which TWO specific corrections should the team apply?

  1. Keep external content in lower-trust user/tool data fields
  2. Split data by appropriate entities and evaluate on unseen cases
  3. Route the request to an image-generation deployment
  4. Evaluate minority-class recall and threshold tradeoffs
  5. Apply clustering to relevant normalized features

Explanation

Correct Answers

B:

Correct for failure (2). A model can memorize training examples without generalizing. It independently corrects the second defect.

E:

Correct for failure (1). Unsupervised learning finds structure without supplied target labels. The proposed correction directly addresses that reported defect.

Incorrect Answers

A:

Instruction roles help distinguish application policy from user task text. The stated pair of failures instead requires “Split data by appropriate entities and evaluate on unseen cases” and “Apply clustering to relevant normalized features”; this option cannot substitute for both.

C:

Text generation and image generation are distinct model output capabilities. The stated pair of failures instead requires “Split data by appropriate entities and evaluate on unseen cases” and “Apply clustering to relevant normalized features”; this option cannot substitute for both.

D:

High accuracy can conceal failure to detect rare positive events. The stated pair of failures instead requires “Split data by appropriate entities and evaluate on unseen cases” and “Apply clustering to relevant normalized features”; this option cannot substitute for both.

Question 468

Troubleshooting report: The UI treats raw unpunctuated tokens as final formatted prose. Which correction is MOST directly targeted to this failure?

  1. Enable supported diarization and validate turn assignments
  2. Apply recognized punctuation support and validate the rendered transcript
  3. Choose the correct TTS voice from the supported catalog
  4. Add the suitable pronunciation markup or lexicon

Explanation

Correct Answer

B:

This correction directly addresses the reported failure. Speech text may require punctuation inference distinct from word recognition. The targeted change is: Apply recognized punctuation support and validate the rendered transcript.

Incorrect Answers

A:

Speaker diarization separates turns and identities at a conversation level. In the stated case, troubleshooting report: The UI treats raw unpunctuated tokens as final formatted prose; “Enable supported diarization and validate turn assignments” does not target that failure.

C:

Voice timbre is a synthesis configuration, not speech recognition output. In the stated case, troubleshooting report: The UI treats raw unpunctuated tokens as final formatted prose; “Choose the correct TTS voice from the supported catalog” does not target that failure.

D:

SSML can guide pronunciation beyond plain-text guessing. In the stated case, troubleshooting report: The UI treats raw unpunctuated tokens as final formatted prose; “Add the suitable pronunciation markup or lexicon” does not target that failure.

Question 469

Troubleshooting report: The system guesses a common road sign based on the question alone. Which correction is MOST directly targeted to this failure?

  1. Attach the image through the documented multimodal input field
  2. Use image input and OCR or vision grounding
  3. Collect balanced labeled examples and test class confusion
  4. Map the task to a whole-image classifier

Explanation

Correct Answer

B:

This correction directly addresses the reported failure. Visual evidence is needed to read sign content. The targeted change is: Use image input and OCR or vision grounding.

Incorrect Answers

A:

A textual filename cannot convey pixels to a model. In the stated case, troubleshooting report: The system guesses a common road sign based on the question alone; “Attach the image through the documented multimodal input field” does not target that failure.

C:

Fine-grained visual classes require representative discriminating images. In the stated case, troubleshooting report: The system guesses a common road sign based on the question alone; “Collect balanced labeled examples and test class confusion” does not target that failure.

D:

Image-level classification is distinct from object localization. In the stated case, troubleshooting report: The system guesses a common road sign based on the question alone; “Map the task to a whole-image classifier” does not target that failure.

Question 470

Troubleshooting report: The application routes the image request to a text-only chat model. Which correction is MOST directly targeted to this failure?

  1. Route the request to an image-generation deployment
  2. Calculate peak token and request demand and request a supported quota or capacity change
  3. Replace it with a suitable text-analysis model or service
  4. Pin compatible deployment settings and maintain versioned evaluations

Explanation

Correct Answer

A:

This correction directly addresses the reported failure. Text generation and image generation are distinct model output capabilities. The targeted change is: Route the request to an image-generation deployment.

Incorrect Answers

B:

Quota sizing is a deployment-capacity decision; retrying a predictable sustained excess cannot add model throughput. In the stated case, troubleshooting report: The application routes the image request to a text-only chat model; “Calculate peak token and request demand and request a supported quota or capacity change” does not target that failure.

C:

Different model tasks require compatible outputs and labeled evaluation data. In the stated case, troubleshooting report: The application routes the image request to a text-only chat model; “Replace it with a suitable text-analysis model or service” does not target that failure.

D:

Reproducibility depends on more than the visible prompt text. In the stated case, troubleshooting report: The application routes the image request to a text-only chat model; “Pin compatible deployment settings and maintain versioned evaluations” does not target that failure.

Microsoft Azure AI Fundamentals (AI-901) Practice Tests