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 2

Question 10

An Azure AI solution uses the following approved control: Pass relevant conversation context or a verified summarized state within token limits. What observable result should pass QA?

  1. A follow-up pronoun resolves to the correct earlier referent
  2. An out-of-scope request is handled under the approved restrictions
  3. An echo-free test response correctly addresses the supplied question
  4. Execution traces show the intended tool invoked with valid arguments

Explanation

Correct Answer

A:

Stateful conversations require supplying or retrieving relevant prior context. The expected acceptance evidence is: A follow-up pronoun resolves to the correct earlier referent. This observation directly tests the stated behavior.

Incorrect Answers

B:

High-priority instructions should describe approved behavior and escalation rules. This observation does not establish the required result: A follow-up pronoun resolves to the correct earlier referent.

C:

The inference service needs task content to generate a relevant answer. This observation does not establish the required result: A follow-up pronoun resolves to the correct earlier referent.

D:

Tools must be discoverable with valid schemas before selection. This observation does not establish the required result: A follow-up pronoun resolves to the correct earlier referent.

Question 11

Troubleshooting report: The system merges both invoices under one vendor record. Which correction is MOST directly targeted to this failure?

  1. Configure supported language handling and test each script
  2. Use schema definitions and grounding evidence for each field
  3. Enable document segmentation and per-document extraction
  4. Use table-aware layout extraction and numeric validation

Explanation

Correct Answer

C:

This correction directly addresses the reported failure. Document boundaries prevent values from unrelated invoices being joined. The targeted change is: Enable document segmentation and per-document extraction.

Incorrect Answers

A:

Recognition accuracy depends on script and language support. In the stated case, troubleshooting report: The system merges both invoices under one vendor record; “Configure supported language handling and test each script” does not target that failure.

B:

Custom extraction maps semantically relevant phrases to typed fields. In the stated case, troubleshooting report: The system merges both invoices under one vendor record; “Use schema definitions and grounding evidence for each field” does not target that failure.

D:

Cell relationships are necessary to check item totals. In the stated case, troubleshooting report: The system merges both invoices under one vendor record; “Use table-aware layout extraction and numeric validation” does not target that failure.

Question 12

A solution team has selected this control: Use managed identity or protected secret storage rather than embedding keys in code. Which observation is the most direct acceptance evidence?

  1. The target model deploys in the approved region and meets residency controls
  2. A valid identity with required resource role succeeds in the test
  3. A repository scan reveals no hardcoded secret and the app still authenticates
  4. Conflict tests keep disallowed actions blocked

Explanation

Correct Answer

C:

Credentials should not appear in repositories, prompt logs or client binaries. The expected acceptance evidence is: A repository scan reveals no hardcoded secret and the app still authenticates. This observation directly tests the stated behavior.

Incorrect Answers

A:

Not every model or feature is available in every location. This observation does not establish the required result: A repository scan reveals no hardcoded secret and the app still authenticates.

B:

Authentication proves caller identity and authorization grants resource access. This observation does not establish the required result: A repository scan reveals no hardcoded secret and the app still authenticates.

D:

Developer and system constraints govern allowed operations over user requests. This observation does not establish the required result: A repository scan reveals no hardcoded secret and the app still authenticates.

Question 13

A solution team has selected this control: Expose uncertainty and route low-confidence image decisions to review. Which observation is the most direct acceptance evidence?

  1. The crop keeps the subject in view at the specified aspect ratio
  2. The output is a validated outdoor probability or category
  3. An ambiguous frame triggers review rather than an irreversible action
  4. The result depicts the prompt without implying original photo preservation

Explanation

Correct Answer

C:

Generated visual interpretations can be wrong even when phrased fluently. The expected acceptance evidence is: An ambiguous frame triggers review rather than an irreversible action. This observation directly tests the stated behavior.

Incorrect Answers

A:

Smart cropping attempts to retain salient regions within target dimensions. This observation does not establish the required result: An ambiguous frame triggers review rather than an irreversible action.

B:

Image-level classification is distinct from object localization. This observation does not establish the required result: An ambiguous frame triggers review rather than an irreversible action.

D:

Generation from text is distinct from image editing. This observation does not establish the required result: An ambiguous frame triggers review rather than an irreversible action.

Question 14

Troubleshooting report: The application calls entity recognition and treats the names as sentiment. Which correction is MOST directly targeted to this failure?

  1. Close the capture stream and update permission controls
  2. Add the suitable pronunciation markup or lexicon
  3. Use the sentiment endpoint and validate its supported language
  4. Use a fixed annotated evaluation set and consistent metrics

Explanation

Correct Answer

C:

This correction directly addresses the reported failure. Sentiment output classifies attitudes expressed in the submitted text. The targeted change is: Use the sentiment endpoint and validate its supported language.

Incorrect Answers

A:

User control of capture must be enforced at the device or application layer. In the stated case, troubleshooting report: The application calls entity recognition and treats the names as sentiment; “Close the capture stream and update permission controls” does not target that failure.

B:

SSML can guide pronunciation beyond plain-text guessing. In the stated case, troubleshooting report: The application calls entity recognition and treats the names as sentiment; “Add the suitable pronunciation markup or lexicon” does not target that failure.

D:

Comparable labeled audio prevents misleading quality claims. In the stated case, troubleshooting report: The application calls entity recognition and treats the names as sentiment; “Use a fixed annotated evaluation set and consistent metrics” does not target that failure.

Question 15

The approved design action is: Run supported image text extraction with field schema and source location. Which test result would BEST establish that it worked?

  1. Each known document type reaches the expected analyzer
  2. The lot code matches the visible label and its image region
  3. The output identifies both amount and confirmed currency
  4. Extracted rows align with the original form cells

Explanation

Correct Answer

B:

Visual information extraction can map pixels to structured identifiers. The expected acceptance evidence is: The lot code matches the visible label and its image region. This observation directly tests the stated behavior.

Incorrect Answers

A:

Different document kinds have distinct field schemas and review paths. This observation does not establish the required result: The lot code matches the visible label and its image region.

C:

A currency symbol or amount alone may not establish financial meaning. This observation does not establish the required result: The lot code matches the visible label and its image region.

D:

Ordinary plain-text OCR may lose table relationships and checkbox states. This observation does not establish the required result: The lot code matches the visible label and its image region.

Question 16

An AI operations team is testing two distinct capabilities. Failure (1): The team assumes fine-tuning always fixes missing source data. Failure (2): The team always selects the largest model without measuring task fit. Which TWO specific corrections should the team apply?

  1. Review label provenance and downstream decision disparities
  2. Run task-specific evaluations before choosing deployment size
  3. Re-embed the corpus or use a versioned compatible index
  4. Replace it with a suitable text-analysis model or service
  5. Separate behavioral model errors from stale or absent knowledge

Explanation

Correct Answers

B:

Correct for failure (2). Model size alone does not establish suitability for a particular task. It independently corrects the second defect.

E:

Correct for failure (1). Model specialization may help task behavior but requires measured benefit. The proposed correction directly addresses that reported defect.

Incorrect Answers

A:

Bias can enter through the targets themselves, not just inputs. In this case the first outcome requires the team to evaluate domain adaptation or fine-tuning only after baseline tests show the gap, and the second requires it to benchmark both on representative quality, latency and cost thresholds. This option does not replace those two separate controls.

C:

Incompatible vector spaces can make nearest-neighbor rankings meaningless. In this case the first outcome requires the team to evaluate domain adaptation or fine-tuning only after baseline tests show the gap, and the second requires it to benchmark both on representative quality, latency and cost thresholds. This option does not replace those two separate controls.

D:

Different model tasks require compatible outputs and labeled evaluation data. In this case the first outcome requires the team to evaluate domain adaptation or fine-tuning only after baseline tests show the gap, and the second requires it to benchmark both on representative quality, latency and cost thresholds. This option does not replace those two separate controls.

Question 17

A solution team has selected this control: Define schema fields for document date, due date and signature date. Which observation is the most direct acceptance evidence?

  1. Speaker assignments align to audible turns with flagged overlaps
  2. The chosen analyzer runs in a supported approved region
  3. Each date field links to the correct labeled region
  4. Mixed-language fields are transcribed correctly on labeled samples

Explanation

Correct Answer

C:

Typed extraction avoids mixing semantically different dates. The expected acceptance evidence is: Each date field links to the correct labeled region. This observation directly tests the stated behavior.

Incorrect Answers

A:

Conversation extraction needs more than a single flattened text stream. This observation does not establish the required result: Each date field links to the correct labeled region.

B:

Some modalities and features depend on resource and API version. This observation does not establish the required result: Each date field links to the correct labeled region.

D:

Recognition accuracy depends on script and language support. This observation does not establish the required result: Each date field links to the correct labeled region.

Question 18

An AI operations team is testing two distinct capabilities. Failure (1): The team always chooses the largest model without task evaluation. Failure (2): A tool executes merely because a user typed please proceed. Which TWO specific corrections should the team apply?

  1. Benchmark candidate deployments and choose by requirements
  2. Move system instructions into protected application configuration
  3. Freeze representative evaluation cases and compare versions
  4. Add approval state and verified role checks
  5. Exercise negative and expected tool paths before release

Explanation

Correct Answers

A:

Correct for failure (1). Lower-cost deployment is acceptable only if accuracy and safety floors remain met. The proposed correction directly addresses that reported defect.

D:

Correct for failure (2). High-impact side effects need permissions independent of conversational language. It independently corrects the second defect.

Incorrect Answers

B:

Versioned system messages provide a controlled behavioral baseline. In this case the first outcome requires the team to choose a model size using representative quality and cost benchmarks, and the second requires it to require explicit confirmation and backend authorization before tool execution. This option does not replace those two separate controls.

C:

Anecdotal success on different prompts cannot establish an improvement. In this case the first outcome requires the team to choose a model size using representative quality and cost benchmarks, and the second requires it to require explicit confirmation and backend authorization before tool execution. This option does not replace those two separate controls.

E:

An answer that sounds plausible may conceal failed tool calls. In this case the first outcome requires the team to choose a model size using representative quality and cost benchmarks, and the second requires it to require explicit confirmation and backend authorization before tool execution. This option does not replace those two separate controls.

Question 19

Troubleshooting report: The team sets fully deterministic output and expects broad variation. Which correction is MOST directly targeted to this failure?

  1. Process appropriate video frames and the audio track
  2. Test a higher diversity setting and compare evaluated outputs
  3. Verify claims against trusted evidence and permit abstention
  4. Choose a recognized audio-input interface and accepted encoding

Explanation

Correct Answer

B:

This correction directly addresses the reported failure. More varied token sampling can increase creative alternatives. The targeted change is: Test a higher diversity setting and compare evaluated outputs.

Incorrect Answers

A:

Different media channels provide complementary scene and dialogue evidence. In the stated case, troubleshooting report: The team sets fully deterministic output and expects broad variation; “Process appropriate video frames and the audio track” does not target that failure.

C:

Language-model likelihood does not independently establish factual truth. In the stated case, troubleshooting report: The team sets fully deterministic output and expects broad variation; “Verify claims against trusted evidence and permit abstention” does not target that failure.

D:

Model capabilities must include the required audio modality before the client can obtain speech-to-text results. In the stated case, troubleshooting report: The team sets fully deterministic output and expects broad variation; “Choose a recognized audio-input interface and accepted encoding” does not target that failure.

Question 20

An Azure-based AI team must satisfy this objective: A support kiosk must respond to spoken input with synthesized speech. Which design choice is best?

  1. Evaluate noise conditions and input quality rather than only clean studio speech
  2. Match the speech resource region or endpoint to the SDK configuration
  3. Apply approved audio retention and access rules and keep only required derived text
  4. Combine recognition or audio understanding with verified text and text-to-speech

Explanation

Correct Answer

D:

Input transcription and output synthesis are distinct steps. This implementation addresses the precise goal rather than an adjacent AI task.

Incorrect Answers

A:

Signal quality directly affects recognizer error rates. That is not sufficient here: Input transcription and output synthesis are distinct steps.

B:

Region or endpoint mismatch prevents requests reaching the correct resource. That is not sufficient here: Input transcription and output synthesis are distinct steps.

C:

Speech workloads still process personal data and require controlled lifecycle. That is not sufficient here: Input transcription and output synthesis are distinct steps.

Microsoft Azure AI Fundamentals (AI-901) Practice Tests