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 78

Question 770

An engineering review documents this requirement: A developer needs to create a chatbot with a persistent corporate system instruction. Select the most appropriate solution.

  1. Archive the actual evaluated response together with the prompt, model deployment, input and generation settings
  2. Store the approved instruction as versioned application configuration
  3. Add an approved knowledge/retrieval tool with clear description and access limits
  4. Limit retrieval and chat context to the authenticated case scope

Explanation

Correct Answer

B:

Versioned system messages provide a controlled behavioral baseline. This implementation addresses the precise goal rather than an adjacent AI task.

Incorrect Answers

A:

Recorded settings support investigation, but generative models can still vary on an apparently identical rerun. That is not sufficient here: Versioned system messages provide a controlled behavioral baseline.

C:

Tool descriptions guide selection, while backend rules enforce allowed sources. That is not sufficient here: Versioned system messages provide a controlled behavioral baseline.

D:

Source scoping enforces privacy and also improves answer relevance. That is not sufficient here: Versioned system messages provide a controlled behavioral baseline.

Question 771

Troubleshooting report: The system outputs a summary instead of a translation. Which correction is MOST directly targeted to this failure?

  1. Use translation and inspect named terms
  2. Run unsupervised clustering and evaluate cluster characteristics
  3. Use a detection output with bounding boxes
  4. Ground the summary in extracted transcript segments

Explanation

Correct Answer

A:

This correction directly addresses the reported failure. Translation transfers meaning across languages rather than changing topic labels. The targeted change is: Use translation and inspect named terms.

Incorrect Answers

B:

Clustering discovers groups based on feature similarity. In the stated case, troubleshooting report: The system outputs a summary instead of a translation; “Run unsupervised clustering and evaluate cluster characteristics” does not target that failure.

C:

Object detectors return classes and spatial boundaries. In the stated case, troubleshooting report: The system outputs a summary instead of a translation; “Use a detection output with bounding boxes” does not target that failure.

D:

Meeting summaries require temporal and spoken source content. In the stated case, troubleshooting report: The system outputs a summary instead of a translation; “Ground the summary in extracted transcript segments” does not target that failure.

Question 772

The approved design action is: Provide the actual image to a capable multimodal or OCR pipeline. Which test result would BEST establish that it worked?

  1. The answer reproduces the visible sign text accurately
  2. The count matches annotated packages after overlap filtering
  3. Restricted users cannot retrieve the original image
  4. A unique feature in the image appears correctly in the response

Explanation

Correct Answer

A:

Visual evidence is needed to read sign content. The expected acceptance evidence is: The answer reproduces the visible sign text accurately. This observation directly tests the stated behavior.

Incorrect Answers

B:

Object counts require resolving distinct instances, not trusting one scene label. This observation does not establish the required result: The answer reproduces the visible sign text accurately.

C:

Images can carry personally identifying information beyond text fields. This observation does not establish the required result: The answer reproduces the visible sign text accurately.

D:

A textual filename cannot convey pixels to a model. This observation does not establish the required result: The answer reproduces the visible sign text accurately.

Question 773

An engineering review documents this requirement: An app sends a picture reference but the model responds with generic text. Select the most appropriate solution.

  1. Use a supported dense-captioning capability where available
  2. Confirm the request includes actual supported image content and an image-capable deployment
  3. Preserve region coordinates with extracted text or field evidence
  4. Use optical character recognition for image text

Explanation

Correct Answer

B:

A textual filename cannot convey pixels to a model. This implementation addresses the precise goal rather than an adjacent AI task.

Incorrect Answers

A:

Dense captions describe localized regions beyond a single global caption. That is not sufficient here: A textual filename cannot convey pixels to a model.

C:

Spatial provenance supports later verification of recognition results. That is not sufficient here: A textual filename cannot convey pixels to a model.

D:

OCR extracts characters visible in pixels. That is not sufficient here: A textual filename cannot convey pixels to a model.

Question 774

An Azure-based AI team must satisfy this objective: A creative writing assistant needs more diverse phrasings than a rigid report generator. Which design choice is best?

  1. Use object detection with bounding boxes and labels
  2. Adjust supported sampling parameters while preserving safety and quality tests
  3. Require verifiable grounding and calibrated uncertainty
  4. Use structured outputs or strict application-side schema validation

Explanation

Correct Answer

B:

More varied token sampling can increase creative alternatives. This implementation addresses the precise goal rather than an adjacent AI task.

Incorrect Answers

A:

An appropriate spatial detection model must localize the defect classes, not merely assign an overall picture label. That is not sufficient here: More varied token sampling can increase creative alternatives.

C:

Language-model likelihood does not independently establish factual truth. That is not sufficient here: More varied token sampling can increase creative alternatives.

D:

Natural-language fluency is not a guarantee of machine-readable contract validity. That is not sufficient here: More varied token sampling can increase creative alternatives.

Question 775

An AI operations team is testing two distinct capabilities. Failure (1): A success message is emitted immediately after invoking an unreliable tool. Failure (2): The team approves deployment based only on a few successful demos. Which TWO specific corrections should the team apply?

  1. Add adversarial and exception-focused cases before promotion
  2. Put an explicit confirmation gate before high-impact actions
  3. Run the frozen adversarial and ordinary acceptance suite against the candidate
  4. Reconcile the provider transaction ID and actual status before confirmation
  5. Create an on-call ownership and escalation schedule

Explanation

Correct Answers

C:

Correct for failure (2). Reliability includes preventing behavior regression across model versions. It independently corrects the second defect.

D:

Correct for failure (1). Generated text is not evidence of external side effects. The proposed correction directly addresses that reported defect.

Incorrect Answers

A:

A high overall score can conceal severe failure on unusual inputs. The stated pair of failures instead requires “Run the frozen adversarial and ordinary acceptance suite against the candidate” and “Reconcile the provider transaction ID and actual status before confirmation”; this option cannot substitute for both.

B:

Reliable assistance distinguishes missing intent from authorized action. It does not complete the two decisions required in this question.

E:

An AI model cannot be its own accountable incident responder. The stated pair of failures instead requires “Run the frozen adversarial and ordinary acceptance suite against the candidate” and “Reconcile the provider transaction ID and actual status before confirmation”; this option cannot substitute for both.

Question 776

An engineering review documents this requirement: A city office wants an AI assistant to write a short response to open-ended citizen inquiries. Select the most appropriate solution.

  1. Use clustering with normalized behavioral features
  2. Use a generative language workload with policy-grounded response constraints
  3. Use image classification for the requested category
  4. Use noise-aware speech recognition and intent classification with fallback

Explanation

Correct Answer

B:

Generation produces novel text, unlike choosing a fixed class label. This implementation addresses the precise goal rather than an adjacent AI task.

Incorrect Answers

A:

Clustering discovers groups based on feature similarity. That is not sufficient here: Generation produces novel text, unlike choosing a fixed class label.

C:

Image classification predicts image-level labels. That is not sufficient here: Generation produces novel text, unlike choosing a fixed class label.

D:

Speech input quality affects subsequent language classification. That is not sufficient here: Generation produces novel text, unlike choosing a fixed class label.

Question 777

Troubleshooting report: The system calls a speech-to-text recognizer with written input. Which correction is MOST directly targeted to this failure?

  1. Invoke a text-to-speech voice and compatible audio format
  2. Supply an authorized identity and matching resource configuration
  3. Set the supported Spanish locale and verify accuracy
  4. Control synthesis prosody in SSML

Explanation

Correct Answer

A:

This correction directly addresses the reported failure. Speech synthesis generates audio waveforms from supplied text. The targeted change is: Invoke a text-to-speech voice and compatible audio format.

Incorrect Answers

B:

Audio capture success does not establish remote service authentication. In the stated case, troubleshooting report: The system calls a speech-to-text recognizer with written input; “Supply an authorized identity and matching resource configuration” does not target that failure.

C:

Recognition models need the appropriate supported language settings. In the stated case, troubleshooting report: The system calls a speech-to-text recognizer with written input; “Set the supported Spanish locale and verify accuracy” does not target that failure.

D:

Voice selection and prosody can influence delivery without changing the text. In the stated case, troubleshooting report: The system calls a speech-to-text recognizer with written input; “Control synthesis prosody in SSML” does not target that failure.

Question 778

An AI operations team is testing two distinct capabilities. Failure (1): The agent copies a different customer's ID from unrelated conversation text. Failure (2): The team tests only whether the chat interface renders text. Which TWO specific corrections should the team apply?

  1. Respect retry guidance and cap concurrent calls
  2. Add idempotency keys and query the original transaction state
  3. Validate case identity and entitlement before each tool call
  4. Exercise negative and expected tool paths before release
  5. Use the supported image-content input and multimodal deployment

Explanation

Correct Answers

C:

Correct for failure (1). Structured state prevents subtle substitution of unrelated customer records. The proposed correction directly addresses that reported defect.

D:

Correct for failure (2). An answer that sounds plausible may conceal failed tool calls. It independently corrects the second defect.

Incorrect Answers

A:

Rate limits protect the service and uncontrolled retries amplify overload. The stated pair of failures instead requires “Validate case identity and entitlement before each tool call” and “Exercise negative and expected tool paths before release”; this option cannot substitute for both.

B:

A timeout does not prove that the original mutation failed. The stated pair of failures instead requires “Validate case identity and entitlement before each tool call” and “Exercise negative and expected tool paths before release”; this option cannot substitute for both.

E:

The model must receive the actual visual content to reason about it. The stated pair of failures instead requires “Validate case identity and entitlement before each tool call” and “Exercise negative and expected tool paths before release”; this option cannot substitute for both.

Question 779

An acceptance tester needs to validate: Implement a no-evidence response instead of fabricating a solution. Which observed result proves the intended behavior?

  1. The response says source evidence is unavailable and suggests safe next steps
  2. Questions without a matching manual passage yield a qualified response
  3. An authenticated test request returns a response from the chosen deployment
  4. A test run shows correct tool selection, arguments and user-visible response

Explanation

Correct Answer

A:

An empty result is not positive evidence for a generated factual claim. The expected acceptance evidence is: The response says source evidence is unavailable and suggests safe next steps. This observation directly tests the stated behavior.

Incorrect Answers

B:

Retrieval access and abstention avoid inventing unsupported documentation. This observation does not establish the required result: The response says source evidence is unavailable and suggests safe next steps.

C:

The portal provides model selection, deployment and test interfaces for supported resources. This observation does not establish the required result: The response says source evidence is unavailable and suggests safe next steps.

D:

An answer that sounds plausible may conceal failed tool calls. This observation does not establish the required result: The response says source evidence is unavailable and suggests safe next steps.

Question 780

An AI operations team is testing two distinct capabilities. Failure (1): A single image classification label cannot identify defect positions. Failure (2): The team calls speech-to-text on already written text. Which TWO specific corrections should the team apply?

  1. Set the API request to the correct deployment name and endpoint
  2. Test a higher diversity setting and compare evaluated outputs
  3. Invoke text-read or document OCR capability
  4. Switch to object detection with coordinate outputs
  5. Use a speech synthesis endpoint and compatible output format

Explanation

Correct Answers

D:

Correct for failure (1). An appropriate spatial detection model must localize the defect classes, not merely assign an overall picture label. The proposed correction directly addresses that reported defect.

E:

Correct for failure (2). Speech synthesis generates audio from text rather than transcribing audio. It independently corrects the second defect.

Incorrect Answers

A:

Applications call a deployed model resource rather than a generic marketing model name. The stated pair of failures instead requires “Switch to object detection with coordinate outputs” and “Use a speech synthesis endpoint and compatible output format”; this option cannot substitute for both.

B:

More varied token sampling can increase creative alternatives. It does not complete the two decisions required in this question.

C:

This action would address another problem: the app returns only visual tags such as paper and desk. OCR produces readable text from pixels. It does not resolve either specified failure.

Microsoft Azure AI Fundamentals (AI-901) Practice Tests