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 48

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.

Question 471

An engineering review documents this requirement: A binary classifier must predict whether a case belongs to a named category. Select the most appropriate solution.

  1. Benchmark both on representative quality, latency and cost thresholds
  2. Use a supervised classification model with representative labeled cases
  3. Use structured outputs or strict application-side schema validation
  4. Choose a speech-recognition model or audio-capable multimodal endpoint with a documented audio-input contract

Explanation

Correct Answer

B:

Classification predicts discrete labels rather than a continuous numeric target. This implementation addresses the precise goal rather than an adjacent AI task.

Incorrect Answers

A:

Model size alone does not establish suitability for a particular task. That is not sufficient here: Classification predicts discrete labels rather than a continuous numeric target.

C:

Natural-language fluency is not a guarantee of machine-readable contract validity. That is not sufficient here: Classification predicts discrete labels rather than a continuous numeric target.

D:

Model capabilities must include the required audio modality before the client can obtain speech-to-text results. That is not sufficient here: Classification predicts discrete labels rather than a continuous numeric target.

Question 472

A project needs the following AI behavior: A research team must condense lengthy research notes into readable highlights. What should the designer implement?

  1. Use visual document extraction with appropriate field and confidence rules
  2. Use text summarization with fidelity checks against sources
  3. Use conversational generative AI with grounded retrieval and scoped dialogue state
  4. Use anomaly detection against expected behavioral patterns

Explanation

Correct Answer

B:

Summarization compresses content without independently validating missing facts. This implementation addresses the precise goal rather than an adjacent AI task.

Incorrect Answers

A:

Images of receipts need OCR/layout extraction before typed business values. That is not sufficient here: Summarization compresses content without independently validating missing facts.

C:

The assistant should incorporate context while remaining connected to reliable sources. That is not sufficient here: Summarization compresses content without independently validating missing facts.

D:

Anomaly methods identify deviations rather than assigning preset incident labels. That is not sufficient here: Summarization compresses content without independently validating missing facts.

Question 473

An Azure AI Fundamentals solution has this requirement: An AI assistant relies on a user assertion of administrator privileges. Which implementation most directly addresses it?

  1. Verify permissions through the identity provider and server-side roles
  2. Require a verified source response before presenting financial values
  3. Enforce the monetary threshold in the backend and require human authorization above it
  4. Measure recognition errors across representative accent and speaking groups

Explanation

Correct Answer

A:

Natural-language claims cannot elevate authentication authority. This implementation addresses the precise goal rather than an adjacent AI task.

Incorrect Answers

B:

A fluent generation cannot substitute for missing authoritative transaction data. That is not sufficient here: Natural-language claims cannot elevate authentication authority.

C:

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

D:

Inclusive voice systems require performance evaluation beyond a single accent. That is not sufficient here: Natural-language claims cannot elevate authentication authority.

Question 474

An AI operations team is testing two distinct capabilities. Failure (1): The app retrains a model to change only section headings. Failure (2): The team compares outputs without noting that the underlying model changed. Which TWO specific corrections should the team apply?

  1. Replace it with a suitable text-analysis model or service
  2. Run task-specific evaluations before choosing deployment size
  3. Pin compatible deployment settings and maintain versioned evaluations
  4. Demonstrate tokenization differences across words and scripts
  5. Start with explicit formatting rules and validate style output

Explanation

Correct Answers

C:

Correct for failure (2). Reproducibility depends on more than the visible prompt text. It independently corrects the second defect.

E:

Correct for failure (1). Clear prompting can control tone without model retraining. The proposed correction directly addresses that reported defect.

Incorrect Answers

A:

Different model tasks require compatible outputs and labeled evaluation data. The stated pair of failures instead requires “Pin compatible deployment settings and maintain versioned evaluations” and “Start with explicit formatting rules and validate style output”; this option cannot substitute for both.

B:

Model size alone does not establish suitability for a particular task. The stated pair of failures instead requires “Pin compatible deployment settings and maintain versioned evaluations” and “Start with explicit formatting rules and validate style output”; this option cannot substitute for both.

D:

Tokenization splits text into subword or other units for model processing. The stated pair of failures instead requires “Pin compatible deployment settings and maintain versioned evaluations” and “Start with explicit formatting rules and validate style output”; this option cannot substitute for both.

Question 475

The approved design action is: Use a held-out evaluation and investigate overfitting or data leakage. Which test result would BEST establish that it worked?

  1. Unseen examples have measured results independent of training
  2. A new approved price appears in answers after source refresh
  3. The returned text includes the photographed serial number
  4. Early stopping selects a model before validation performance degrades

Explanation

Correct Answer

A:

A model can memorize training examples without generalizing. The expected acceptance evidence is: Unseen examples have measured results independent of training. This observation directly tests the stated behavior.

Incorrect Answers

B:

External retrieval updates factual context without retraining model weights. This observation does not establish the required result: Unseen examples have measured results independent of training.

C:

OCR produces readable text from pixels. This observation does not establish the required result: Unseen examples have measured results independent of training.

D:

More optimization on training data can worsen generalization. This observation does not establish the required result: Unseen examples have measured results independent of training.

Question 476

The approved design action is: Choose media analyzers with transcript and scene evidence tied to the retrieved answer. Which test result would BEST establish that it worked?

  1. Each date field links to the correct labeled region
  2. A reply links its claim to a transcript time or relevant video segment
  3. A question links to the correct contract section and page
  4. The lot code matches the visible label and its image region

Explanation

Correct Answer

B:

Multimodal grounding should preserve how the answer maps to source media. The expected acceptance evidence is: A reply links its claim to a transcript time or relevant video segment. This observation directly tests the stated behavior.

Incorrect Answers

A:

Typed extraction avoids mixing semantically different dates. This observation does not establish the required result: A reply links its claim to a transcript time or relevant video segment.

C:

Structured layout and provenance let users verify a quoted clause. This observation does not establish the required result: A reply links its claim to a transcript time or relevant video segment.

D:

Visual information extraction can map pixels to structured identifiers. This observation does not establish the required result: A reply links its claim to a transcript time or relevant video segment.

Question 477

Troubleshooting report: The tool exists in backend code but is not attached to the agent. Which correction is MOST directly targeted to this failure?

  1. Rewrite the prompt with explicit acceptance criteria
  2. Require real tool calls and distinguish pending from completed
  3. Register the tool and validate its description and argument schema
  4. Reinforce role boundaries and enforce secrets outside prompts

Explanation

Correct Answer

C:

This correction directly addresses the reported failure. Tools must be discoverable with valid schemas before selection. The targeted change is: Register the tool and validate its description and argument schema.

Incorrect Answers

A:

Specific instructions reduce ambiguity about what constitutes success. In the stated case, troubleshooting report: The tool exists in backend code but is not attached to the agent; “Rewrite the prompt with explicit acceptance criteria” does not target that failure.

B:

Generated text does not prove external execution. In the stated case, troubleshooting report: The tool exists in backend code but is not attached to the agent; “Require real tool calls and distinguish pending from completed” does not target that failure.

D:

Developer and system constraints govern allowed operations over user requests. In the stated case, troubleshooting report: The tool exists in backend code but is not attached to the agent; “Reinforce role boundaries and enforce secrets outside prompts” does not target that failure.

Question 478

A project needs the following AI behavior: A quality-control service receives pictures with incorrect orientation. What should the designer implement?

  1. Use an image-generation-capable model with an appropriate prompt
  2. Normalize image orientation before applying vision detection
  3. Validate file type, size and image dimensions before the model call
  4. Use optical character recognition for image text

Explanation

Correct Answer

B:

Rotated inputs can degrade detection or coordinate interpretation. This implementation addresses the precise goal rather than an adjacent AI task.

Incorrect Answers

A:

Image generation creates new pixels rather than classifying an existing photo. That is not sufficient here: Rotated inputs can degrade detection or coordinate interpretation.

C:

Vision endpoints impose supported input formats and limits. That is not sufficient here: Rotated inputs can degrade detection or coordinate interpretation.

D:

OCR extracts characters visible in pixels. That is not sufficient here: Rotated inputs can degrade detection or coordinate interpretation.

Question 479

Troubleshooting report: Only a still image is provided to summarize an entire conversation. Which correction is MOST directly targeted to this failure?

  1. Invoke text-read or document OCR capability
  2. Reduce sampling variability and validate outputs against a schema
  3. Validate the tool call and observe an actual execution receipt
  4. Process appropriate video frames and the audio track

Explanation

Correct Answer

D:

This correction directly addresses the reported failure. Different media channels provide complementary scene and dialogue evidence. The targeted change is: Process appropriate video frames and the audio track.

Incorrect Answers

A:

OCR produces readable text from pixels. In the stated case, troubleshooting report: Only a still image is provided to summarize an entire conversation; “Invoke text-read or document OCR capability” does not target that failure.

B:

Lower randomness often improves consistency but does not guarantee correctness. In the stated case, troubleshooting report: Only a still image is provided to summarize an entire conversation; “Reduce sampling variability and validate outputs against a schema” does not target that failure.

C:

Agent workflows can reason about tool selection but tools still need authorization. In the stated case, troubleshooting report: Only a still image is provided to summarize an entire conversation; “Validate the tool call and observe an actual execution receipt” does not target that failure.

Question 480

An Azure AI solution uses the following approved control: Wait for supported final recognition results before high-impact tool calls. What observable result should pass QA?

  1. Subtitle cues match the spoken phrases at measured times
  2. Noise-augmented recognition tests meet requirements or request repetition
  3. A confirmed final transcript triggers the intended action once
  4. The listener hears equivalent French instructions

Explanation

Correct Answer

C:

Interim hypotheses can change as more audio arrives. The expected acceptance evidence is: A confirmed final transcript triggers the intended action once. This observation directly tests the stated behavior.

Incorrect Answers

A:

Timecodes align text segments to audio playback. This observation does not establish the required result: A confirmed final transcript triggers the intended action once.

B:

Signal quality directly affects recognizer error rates. This observation does not establish the required result: A confirmed final transcript triggers the intended action once.

D:

Cross-language audio output requires both language transfer and synthesis. This observation does not establish the required result: A confirmed final transcript triggers the intended action once.

Microsoft Azure AI Fundamentals (AI-901) Practice Tests