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 63

Question 620

Troubleshooting report: Evaluation uses only studio-recorded standard-accent speech. Which correction is MOST directly targeted to this failure?

  1. Enforce backend authorization and signed approval regardless of prompt wording
  2. Expand the consented voice test corpus and provide fallback interaction
  3. Add low-light and motion-blur scenarios to the evaluation set
  4. Propagate typed tool errors and use an explicit unavailable response

Explanation

Correct Answer

B:

This correction directly addresses the reported failure. Inclusive voice systems require performance evaluation beyond a single accent. The targeted change is: Expand the consented voice test corpus and provide fallback interaction.

Incorrect Answers

A:

Accountable high-impact operations need verifiable authorized signatures. In the stated case, troubleshooting report: Evaluation uses only studio-recorded standard-accent speech; “Enforce backend authorization and signed approval regardless of prompt wording” does not target that failure.

C:

Reliability demands testing the actual operating environment, not just clean samples. In the stated case, troubleshooting report: Evaluation uses only studio-recorded standard-accent speech; “Add low-light and motion-blur scenarios to the evaluation set” does not target that failure.

D:

A fluent generation cannot substitute for missing authoritative transaction data. In the stated case, troubleshooting report: Evaluation uses only studio-recorded standard-accent speech; “Propagate typed tool errors and use an explicit unavailable response” does not target that failure.

Question 621

Troubleshooting report: The team compares outputs without noting that the underlying model changed. Which correction is MOST directly targeted to this failure?

  1. Demonstrate probabilistic token generation with contrasting prompts
  2. Pin compatible deployment settings and maintain versioned evaluations
  3. Verify claims against trusted evidence and permit abstention
  4. Run task-specific evaluations before choosing deployment size

Explanation

Correct Answer

B:

This correction directly addresses the reported failure. Reproducibility depends on more than the visible prompt text. The targeted change is: Pin compatible deployment settings and maintain versioned evaluations.

Incorrect Answers

A:

A generative language model estimates likely continuations rather than reading a stored answer paragraph. In the stated case, troubleshooting report: The team compares outputs without noting that the underlying model changed; “Demonstrate probabilistic token generation with contrasting prompts” does not target that failure.

C:

Language-model likelihood does not independently establish factual truth. In the stated case, troubleshooting report: The team compares outputs without noting that the underlying model changed; “Verify claims against trusted evidence and permit abstention” does not target that failure.

D:

Model size alone does not establish suitability for a particular task. In the stated case, troubleshooting report: The team compares outputs without noting that the underlying model changed; “Run task-specific evaluations before choosing deployment size” does not target that failure.

Question 622

Troubleshooting report: The agent blindly invokes create-order again after a timeout. Which correction is MOST directly targeted to this failure?

  1. Provision proper identity and resource role bindings for the target tenant
  2. Isolate request state and bind responses to their originating session
  3. Add idempotency keys and query the original transaction state
  4. Reinforce role boundaries and enforce secrets outside prompts

Explanation

Correct Answer

C:

This correction directly addresses the reported failure. A timeout does not prove that the original mutation failed. The targeted change is: Add idempotency keys and query the original transaction state.

Incorrect Answers

A:

Copying code does not grant access to another tenant's resources. In the stated case, troubleshooting report: The agent blindly invokes create-order again after a timeout; “Provision proper identity and resource role bindings for the target tenant” does not target that failure.

B:

Concurrent inference must not mix user data or response ownership. In the stated case, troubleshooting report: The agent blindly invokes create-order again after a timeout; “Isolate request state and bind responses to their originating session” does not target that failure.

D:

Developer and system constraints govern allowed operations over user requests. In the stated case, troubleshooting report: The agent blindly invokes create-order again after a timeout; “Reinforce role boundaries and enforce secrets outside prompts” does not target that failure.

Question 623

The approved design action is: Capture allowed tool calls, response statuses and source references in traces. Which test result would BEST establish that it worked?

  1. The agent selects the expected tool for a supported test query
  2. An account-specific summary excludes cross-account conversation data
  3. Known unsupported claims remain flagged regardless of temperature
  4. A trace shows which tool result supported each claimed business action

Explanation

Correct Answer

D:

Operational observability aids debugging without exposing hidden sensitive prompt data. The expected acceptance evidence is: A trace shows which tool result supported each claimed business action. This observation directly tests the stated behavior.

Incorrect Answers

A:

A single agent can orchestrate a bounded tool without a multi-agent system. This observation does not establish the required result: A trace shows which tool result supported each claimed business action.

B:

Source scoping enforces privacy and also improves answer relevance. This observation does not establish the required result: A trace shows which tool result supported each claimed business action.

C:

Sampling parameters influence diversity but do not verify knowledge. This observation does not establish the required result: A trace shows which tool result supported each claimed business action.

Question 624

Troubleshooting report: The agent can call a payment API directly with no server-side limit. Which correction is MOST directly targeted to this failure?

  1. Exclude obsolete material with approval and timestamp filters
  2. Gate the payment tool with verified role and amount checks
  3. Apply real sign-in and role checks before invoking protected tools
  4. Permit calibrated caveats and qualified review paths

Explanation

Correct Answer

B:

This correction directly addresses the reported failure. A prompt-level request cannot override a transaction policy. The targeted change is: Gate the payment tool with verified role and amount checks.

Incorrect Answers

A:

A precise explanation of an obsolete procedure can still be dangerous. In the stated case, troubleshooting report: The agent can call a payment API directly with no server-side limit; “Exclude obsolete material with approval and timestamp filters” does not target that failure.

C:

Natural-language claims cannot elevate authentication authority. In the stated case, troubleshooting report: The agent can call a payment API directly with no server-side limit; “Apply real sign-in and role checks before invoking protected tools” does not target that failure.

D:

Transparency reduces mistaken reliance on unsupported professional guarantees. In the stated case, troubleshooting report: The agent can call a payment API directly with no server-side limit; “Permit calibrated caveats and qualified review paths” does not target that failure.

Question 625

An AI operations team is testing two distinct capabilities. Failure (1): The training set contains only one defect class. Failure (2): Overlapping detections are counted twice. Which TWO specific corrections should the team apply?

  1. Collect balanced labeled examples and test class confusion
  2. Apply confidence and nonmaximum suppression or instance deduplication
  3. Limit concurrency and honor retry guidance
  4. Gate distribution on the image moderation result
  5. Request tags and parse the tag collection

Explanation

Correct Answers

A:

Correct for failure (1). Fine-grained visual classes require representative discriminating images. The proposed correction directly addresses that reported defect.

B:

Correct for failure (2). Object counts require resolving distinct instances, not trusting one scene label. It independently corrects the second defect.

Incorrect Answers

C:

Too many concurrent requests can exceed resource quota. It does not complete the two decisions required in this question.

D:

Generative media needs policy controls in addition to prompt instructions. The stated pair of failures instead requires “Collect balanced labeled examples and test class confusion” and “Apply confidence and nonmaximum suppression or instance deduplication”; 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 626

The approved design action is: Escape XML special characters and validate SSML syntax. Which test result would BEST establish that it worked?

  1. The sanitized SSML parses and renders the intended characters
  2. Each action item maps to an utterance and assigned speaker
  3. The output audio reads the approved sentence intelligibly
  4. A labeled command yields readable sentence boundaries

Explanation

Correct Answer

A:

SSML is XML and malformed markup breaks the synthesis request. The expected acceptance evidence is: The sanitized SSML parses and renders the intended characters. This observation directly tests the stated behavior.

Incorrect Answers

B:

Transcription obtains words; summarization extracts decisions and tasks. This observation does not establish the required result: The sanitized SSML parses and renders the intended characters.

C:

Speech synthesis generates audio waveforms from supplied text. This observation does not establish the required result: The sanitized SSML parses and renders the intended characters.

D:

Speech text may require punctuation inference distinct from word recognition. This observation does not establish the required result: The sanitized SSML parses and renders the intended characters.

Question 627

An AI operations team is testing two distinct capabilities. Failure (1): The synth result is saved but no playback channel is opened. Failure (2): The interface says stopped while background capture continues. Which TWO specific corrections should the team apply?

  1. Configure playback routing and compatible output decoding
  2. Correct streaming buffer and session lifecycle
  3. Adjust recognition vocabulary and test the target terms
  4. Close the capture stream and update permission controls
  5. Enable supported language detection with constrained candidates

Explanation

Correct Answers

A:

Correct for failure (1). Successful synthesis does not guarantee local audio is routed to a speaker. The proposed correction directly addresses that reported defect.

D:

Correct for failure (2). User control of capture must be enforced at the device or application layer. It independently corrects the second defect.

Incorrect Answers

B:

Real-time audio pipelines need coordinated capture and recognition start. The stated pair of failures instead requires “Configure playback routing and compatible output decoding” and “Close the capture stream and update permission controls”; this option cannot substitute for both.

C:

Domain-specific words may require biasing or adaptation beyond a generic recognizer. The stated pair of failures instead requires “Configure playback routing and compatible output decoding” and “Close the capture stream and update permission controls”; this option cannot substitute for both.

E:

Automatic language detection chooses among an approved set of languages. The stated pair of failures instead requires “Configure playback routing and compatible output decoding” and “Close the capture stream and update permission controls”; this option cannot substitute for both.

Question 628

A solution team has selected this control: Choose a model or pipeline that supports both visual and audio modalities. Which observation is the most direct acceptance evidence?

  1. The returned text includes the photographed serial number
  2. The holdout report identifies missed positives and false alarms explicitly
  3. The summary references events visible on frames and spoken in the transcript
  4. Queries and indexed vectors carry matching model-version metadata

Explanation

Correct Answer

C:

Different media channels provide complementary scene and dialogue evidence. The expected acceptance evidence is: The summary references events visible on frames and spoken in the transcript. 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 summary references events visible on frames and spoken in the transcript.

B:

High accuracy can conceal failure to detect rare positive events. This observation does not establish the required result: The summary references events visible on frames and spoken in the transcript.

D:

Incompatible vector spaces can make nearest-neighbor rankings meaningless. This observation does not establish the required result: The summary references events visible on frames and spoken in the transcript.

Question 629

A project needs the following AI behavior: A team wants to generate code snippets from a natural-language feature request. What should the designer implement?

  1. Put persistent task boundaries in a supported high-priority system instruction
  2. Use a compatible code-capable generative model with constrained context and tests
  3. Archive the actual evaluated response together with the prompt, model deployment, input and generation settings
  4. Write an instruction to identify missing inputs and request clarification before tools

Explanation

Correct Answer

B:

Generated code must be validated, not treated as production-ready from fluency. This implementation addresses the precise goal rather than an adjacent AI task.

Incorrect Answers

A:

High-priority instructions should describe approved behavior and escalation rules. That is not sufficient here: Generated code must be validated, not treated as production-ready from fluency.

C:

Recorded settings support investigation, but generative models can still vary on an apparently identical rerun. That is not sufficient here: Generated code must be validated, not treated as production-ready from fluency.

D:

Prompts can define required preconditions but tools must still enforce them. That is not sufficient here: Generated code must be validated, not treated as production-ready from fluency.

Question 630

An acceptance tester needs to validate: Use the box coordinates to locate detected people in the source image. Which observed result proves the intended behavior?

  1. A drawn overlay aligns each box with the corresponding person
  2. The result includes supported text, tags and person boxes without conflation
  3. The product outline remains intact after background editing
  4. The summary preserves the order of observed demonstration steps and identifies gaps or uncertain actions

Explanation

Correct Answer

A:

Localization metadata is tied to pixel regions, not just labels. The expected acceptance evidence is: A drawn overlay aligns each box with the corresponding person. This observation directly tests the stated behavior.

Incorrect Answers

B:

Different visual outputs describe text, scene classes and detected people. This observation does not establish the required result: A drawn overlay aligns each box with the corresponding person.

C:

A mask defines which image area may change. This observation does not establish the required result: A drawn overlay aligns each box with the corresponding person.

D:

A single still cannot convey action order; sampled video frames may also omit fleeting intermediate steps. This observation does not establish the required result: A drawn overlay aligns each box with the corresponding person.

Microsoft Azure AI Fundamentals (AI-901) Practice Tests