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 12

Question 110

Troubleshooting report: Every upload is forced through the invoice schema. Which correction is MOST directly targeted to this failure?

  1. Use entity recognition and verify sensitive-entity recall
  2. Call the authorized ticket tool and verify the receipt
  3. Introduce a document classifier and route by type
  4. Enforce backend identity and method scopes

Explanation

Correct Answer

C:

This correction directly addresses the reported failure. Classification selects the extraction process matching the document category. The targeted change is: Introduce a document classifier and route by type.

Incorrect Answers

A:

Entity extraction locates sensitive spans for privacy handling. In the stated case, troubleshooting report: Every upload is forced through the invoice schema; “Use entity recognition and verify sensitive-entity recall” does not target that failure.

B:

Agentic workloads combine reasoning and bounded external actions. In the stated case, troubleshooting report: Every upload is forced through the invoice schema; “Call the authorized ticket tool and verify the receipt” does not target that failure.

D:

Agentic workflows combine information access with controlled actions. In the stated case, troubleshooting report: Every upload is forced through the invoice schema; “Enforce backend identity and method scopes” does not target that failure.

Question 111

Troubleshooting report: The sizing plan assumes unlimited requests and tokens despite a documented fixed quota. Which correction is MOST directly targeted to this failure?

  1. Calculate peak token and request demand and request a supported quota or capacity change
  2. Use a multimodal-capable deployment and supported message format
  3. Run task-specific evaluations before choosing deployment size
  4. Use a speech synthesis endpoint and compatible output format

Explanation

Correct Answer

A:

This correction directly addresses the reported failure. Quota sizing is a deployment-capacity decision; retrying a predictable sustained excess cannot add model throughput. The targeted change is: Calculate peak token and request demand and request a supported quota or capacity change.

Incorrect Answers

B:

Multimodal models can reason over more than one supported input modality. In the stated case, troubleshooting report: The sizing plan assumes unlimited requests and tokens despite a documented fixed quota; “Use a multimodal-capable deployment and supported message format” does not target that failure.

C:

Model size alone does not establish suitability for a particular task. In the stated case, troubleshooting report: The sizing plan assumes unlimited requests and tokens despite a documented fixed quota; “Run task-specific evaluations before choosing deployment size” does not target that failure.

D:

Speech synthesis generates audio from text rather than transcribing audio. In the stated case, troubleshooting report: The sizing plan assumes unlimited requests and tokens despite a documented fixed quota; “Use a speech synthesis endpoint and compatible output format” does not target that failure.

Question 112

An AI operations team is testing two distinct capabilities. Failure (1): The model states a ticket exists without invoking the ticket API. Failure (2): The system reports only high/low categories without quantities. Which TWO specific corrections should the team apply?

  1. Select visual sign recognition rather than text-only extraction
  2. Use time-series forecasting and numeric error metrics
  3. Call the authorized ticket tool and verify the receipt
  4. Add retrieval and grounded citation generation
  5. Run unsupervised clustering and evaluate cluster characteristics

Explanation

Correct Answers

B:

Correct for failure (2). Forecasts predict future numeric quantities rather than discrete tags. It independently corrects the second defect.

C:

Correct for failure (1). Agentic workloads combine reasoning and bounded external actions. The proposed correction directly addresses that reported defect.

Incorrect Answers

A:

The task chooses a label, with localization needed only when multiple signs appear. In this case the first outcome requires the team to use an agent workflow with a permitted ticket tool and an authorization gate, and the second requires it to use regression or forecasting appropriate to time-series data. This option does not replace those two separate controls.

D:

RAG retrieves current external evidence for a generated response. In this case the first outcome requires the team to use an agent workflow with a permitted ticket tool and an authorization gate, and the second requires it to use regression or forecasting appropriate to time-series data. This option does not replace those two separate controls.

E:

Clustering discovers groups based on feature similarity. In this case the first outcome requires the team to use an agent workflow with a permitted ticket tool and an authorization gate, and the second requires it to use regression or forecasting appropriate to time-series data. This option does not replace those two separate controls.

Question 113

Troubleshooting report: The service returns coordinates but no document-level label. Which correction is MOST directly targeted to this failure?

  1. Replace it with a suitable text-analysis model or service
  2. Select a classification output for image-wide categories
  3. Use a speech synthesis endpoint and compatible output format
  4. Define allowed JSON keys and validate the returned types

Explanation

Correct Answer

B:

This correction directly addresses the reported failure. Classification assigns image-wide categories rather than bounding-box coordinates. The targeted change is: Select a classification output for image-wide categories.

Incorrect Answers

A:

Different model tasks require compatible outputs and labeled evaluation data. In the stated case, troubleshooting report: The service returns coordinates but no document-level label; “Replace it with a suitable text-analysis model or service” does not target that failure.

C:

Speech synthesis generates audio from text rather than transcribing audio. In the stated case, troubleshooting report: The service returns coordinates but no document-level label; “Use a speech synthesis endpoint and compatible output format” does not target that failure.

D:

Natural-language fluency is not a guarantee of machine-readable contract validity. In the stated case, troubleshooting report: The service returns coordinates but no document-level label; “Define allowed JSON keys and validate the returned types” does not target that failure.

Question 114

An Azure-based AI team must satisfy this objective: A customs form includes many repeated line items. Which design choice is best?

  1. Use array-typed field extraction rather than one scalar field
  2. Use document extraction with source positions before indexing for search
  3. Use timestamped evidence-linked video segments and explicitly account for frame-sampling limits
  4. Use an appropriate search-oriented document analyzer with meaningful chunk boundaries

Explanation

Correct Answer

A:

Repeated records require structured lists with item-level associations. This implementation addresses the precise goal rather than an adjacent AI task.

Incorrect Answers

B:

Structured layout and provenance let users verify a quoted clause. That is not sufficient here: Repeated records require structured lists with item-level associations.

C:

Temporal evidence supports the timing of observed events, but approximately one frame per second cannot guarantee subsecond event capture. That is not sufficient here: Repeated records require structured lists with item-level associations.

D:

RAG preparation should preserve document structure and source references. That is not sufficient here: Repeated records require structured lists with item-level associations.

Question 115

The approved design action is: Check output-token limits and context-budget allocation. Which test result would BEST establish that it worked?

  1. A permitted larger output budget allows the complete response
  2. Repeated structured cases follow the expected schema more consistently
  3. The output is produced as a sequence of tokens conditioned on the prompt
  4. Held-out defect images meet the tested localization and per-class detection thresholds

Explanation

Correct Answer

A:

Generation stops when an output budget or stop condition is reached. The expected acceptance evidence is: A permitted larger output budget allows the complete response. This observation directly tests the stated behavior.

Incorrect Answers

B:

Lower randomness often improves consistency but does not guarantee correctness. This observation does not establish the required result: A permitted larger output budget allows the complete response.

C:

A generative language model estimates likely continuations rather than reading a stored answer paragraph. This observation does not establish the required result: A permitted larger output budget allows the complete response.

D:

An appropriate spatial detection model must localize the defect classes, not merely assign an overall picture label. This observation does not establish the required result: A permitted larger output budget allows the complete response.

Question 116

An acceptance tester needs to validate: Segment authorized audio by task and retention purpose before processing. Which observed result proves the intended behavior?

  1. The narration pauses at marked section boundaries
  2. Only the needed audio excerpts are sent for transcription
  3. The spoken answer slows while preserving its content
  4. Subtitle cues match the spoken phrases at measured times

Explanation

Correct Answer

B:

Minimization reduces unnecessary personal-data exposure. The expected acceptance evidence is: Only the needed audio excerpts are sent for transcription. This observation directly tests the stated behavior.

Incorrect Answers

A:

SSML controls aspects of prosody and timing for synthesized speech. This observation does not establish the required result: Only the needed audio excerpts are sent for transcription.

C:

Voice selection and prosody can influence delivery without changing the text. This observation does not establish the required result: Only the needed audio excerpts are sent for transcription.

D:

Timecodes align text segments to audio playback. This observation does not establish the required result: Only the needed audio excerpts are sent for transcription.

Question 117

A solution team has selected this control: Use speech-to-text recognition on a supported audio input stream. Which observation is the most direct acceptance evidence?

  1. Early spoken words are present in a captured test transcript
  2. The transcript matches a labeled spoken test phrase
  3. A confirmed final transcript triggers the intended action once
  4. The model's response reflects the words spoken in the actual recording

Explanation

Correct Answer

B:

Speech recognition converts waveform features to text tokens. The expected acceptance evidence is: The transcript matches a labeled spoken test phrase. This observation directly tests the stated behavior.

Incorrect Answers

A:

Real-time audio pipelines need coordinated capture and recognition start. This observation does not establish the required result: The transcript matches a labeled spoken test phrase.

C:

Interim hypotheses can change as more audio arrives. This observation does not establish the required result: The transcript matches a labeled spoken test phrase.

D:

A text-only request cannot interpret sound content by file name alone. This observation does not establish the required result: The transcript matches a labeled spoken test phrase.

Question 118

An engineering review documents this requirement: A speech SDK returns an authentication error even though the microphone works. Select the most appropriate solution.

  1. Use text summarization with clear length and fidelity constraints
  2. Use a supported text-to-speech synthesis deployment
  3. Check playback device selection and the synthesized audio format
  4. Verify the service credential, resource and authorized region

Explanation

Correct Answer

D:

Audio capture success does not establish remote service authentication. This implementation addresses the precise goal rather than an adjacent AI task.

Incorrect Answers

A:

Summarization compresses written information rather than changing its facts. That is not sufficient here: Audio capture success does not establish remote service authentication.

B:

Speech synthesis generates audio waveforms from supplied text. That is not sufficient here: Audio capture success does not establish remote service authentication.

C:

Successful synthesis does not guarantee local audio is routed to a speaker. That is not sufficient here: Audio capture success does not establish remote service authentication.

Question 119

A project needs the following AI behavior: An automated support agent may issue refunds above a company approval limit. What should the designer implement?

  1. Verify permissions through the identity provider and server-side roles
  2. Enforce the monetary threshold in the backend and require human authorization above it
  3. Use grounded medical content, high-risk detection and appropriate expert escalation
  4. Exclude protected proxies from eligibility logic and test the remaining permitted factors

Explanation

Correct Answer

B:

A prompt-level request cannot override a transaction policy. This implementation addresses the precise goal rather than an adjacent AI task.

Incorrect Answers

A:

Natural-language claims cannot elevate authentication authority. That is not sufficient here: A prompt-level request cannot override a transaction policy.

C:

Safety mechanisms should reduce unsupported high-impact recommendations. That is not sufficient here: A prompt-level request cannot override a transaction policy.

D:

Legitimate credit criteria cannot be replaced by sensitive identity traits. That is not sufficient here: A prompt-level request cannot override a transaction policy.

Question 120

A project needs the following AI behavior: A user requests a sensitive irreversible action through an agent. What should the designer implement?

  1. Require explicit confirmation and backend authorization before tool execution
  2. Write an instruction to identify missing inputs and request clarification before tools
  3. Select and test a compatible deployed model through the Microsoft Foundry portal
  4. Use a versioned evaluation set with stable scoring and test inputs

Explanation

Correct Answer

A:

High-impact side effects need permissions independent of conversational language. This implementation addresses the precise goal rather than an adjacent AI task.

Incorrect Answers

B:

Prompts can define required preconditions but tools must still enforce them. That is not sufficient here: High-impact side effects need permissions independent of conversational language.

C:

The portal provides model selection, deployment and test interfaces for supported resources. That is not sufficient here: High-impact side effects need permissions independent of conversational language.

D:

Anecdotal success on different prompts cannot establish an improvement. That is not sufficient here: High-impact side effects need permissions independent of conversational language.

Microsoft Azure AI Fundamentals (AI-901) Practice Tests