Practice Exams:

Microsoft AI-103 Dumps

EXAM: Developing AI Apps and Agents on Azure
Verified By Experts
Premium Bundle
$49.99

AI-103 Premium ETE File


  • Premium File 187 Questions & Answers
Whats Included:
  • Latest Questions
  • 100% Accurate Answers
  • Fast Exam Updates

Microsoft AI-103 Practice Test 66

Question 650

During a service audit, a cloud developer must accept a change for cost and quality dashboard. The issue to be addressed is: Operations cannot tell whether a model change saved money without hurting task success. Which concrete verification action demonstrates that the relevant behavior works?

  1. Run matched traffic samples and compare costs and ground-truth outcomes
  2. Report token savings alone without comparing answer accuracy
  3. Track per-intent quality token cost throughput and latency together
  4. Repeat fixed classification inputs and compare consistency and accuracy

Explanation

Correct Answer

A:

To verify this change, the industrial maintenance team should run matched traffic samples and compare costs and ground-truth outcomes. Operators still need the implementation to track per-intent quality token cost throughput and latency together and the preventive requirement to define joint cost-quality acceptance criteria and baseline cohorts.

Incorrect Answers

B:

In the industrial maintenance team acceptance review, trying to report token savings alone without comparing answer accuracy does not test the exact cost and quality dashboard condition where operations cannot tell whether a model change saved money without hurting task success; the relevant test is to run matched traffic samples and compare costs and ground-truth outcomes. Acceptance should instead demonstrate that the service can run matched traffic samples and compare costs and ground-truth outcomes, using the described failure as the test criterion.

C:

The action to track per-intent quality token cost throughput and latency together can contribute to cost and quality dashboard as a implementation step; here, operations cannot tell whether a model change saved money without hurting task success. In the industrial maintenance team acceptance review, trying to track per-intent quality token cost throughput and latency together does not test the exact cost and quality dashboard condition where operations cannot tell whether a model change saved money without hurting task success; the relevant test is to run matched traffic samples and compare costs and ground-truth outcomes.

D:

For temperature tuning, it can be sensible to repeat fixed classification inputs and compare consistency and accuracy if the same billing input produces materially different classifications across runs. Without the actual change to run matched traffic samples and compare costs and ground-truth outcomes, the answer lacks the correct operational scope even if the alternate action is technically valid elsewhere.

Question 651

A site reliability engineer at a transport authority is conducting a release acceptance review concerning text-driven image generation. The observed problem is: A design team must generate an original illustration from a natural-language description. Which TWO actions together represent the needed implementation AND a durable preventive safeguard? (Select TWO answers.)

  1. Assume a chat-only text endpoint automatically returns binary image assets
  2. Submit a sample prompt and inspect generated image dimensions and relevance
  3. Use a supported prompt-based editing capability with preservation instructions
  4. Check permitted content and output asset type before rendering
  5. Use a supported image-generation model with a detailed permitted text prompt

Explanation

Correct Answers

D:

The sustained prevention measure is to check permitted content and output asset type before rendering. The implementation can still use the separate action to use a supported image-generation model with a detailed permitted text prompt, but a guardrail constrains future changes. The transport authority can measure whether the rule works by arranging to submit a sample prompt and inspect generated image dimensions and relevance.

E:

The proposed implementation is to use a supported image-generation model with a detailed permitted text prompt; this matches the operational requirement that a design team must generate an original illustration from a natural-language description. The team can submit a sample prompt and inspect generated image dimensions and relevance. The team should also check permitted content and output asset type before rendering.

Incorrect Answers

A:

Choosing to assume a chat-only text endpoint automatically returns binary image assets cannot directly replace the specified action to check permitted content and output asset type before rendering when a design team must generate an original illustration from a natural-language description. The actual implementation must use a supported image-generation model with a detailed permitted text prompt, because the stated service condition is that a design team must generate an original illustration from a natural-language description.

B:

The action to submit a sample prompt and inspect generated image dimensions and relevance can contribute to text-driven image generation as a targeted acceptance check step; here, a design team must generate an original illustration from a natural-language description. Choosing to submit a sample prompt and inspect generated image dimensions and relevance cannot directly replace the specified action to check permitted content and output asset type before rendering when a design team must generate an original illustration from a natural-language description.

C:

The option to use a supported prompt-based editing capability with preservation instructions would normally serve a different prompt-driven image edit requirement, where a user requests a changed sky while retaining architecture and perspective. The alternative may be useful in its own context, but it does not directly supply use a supported image-generation model with a detailed permitted text prompt for the stated decision.

Question 652

After a design meeting at a large online retailer, the team observes this failure: A design team must generate an original illustration from a natural-language description. Which flawed assumption or implementation choice would MOST DIRECTLY explain why the intended outcome was not achieved?

  1. Assume all visual model endpoints accept arbitrary pixel sizes
  2. Assume a chat-only text endpoint automatically returns binary image assets
  3. Use a supported image-generation model with a detailed permitted text prompt
  4. Submit a sample prompt and inspect generated image dimensions and relevance

Explanation

Correct Answer

B:

This design mistake is to assume a chat-only text endpoint automatically returns binary image assets; it conflicts with the actual text-driven image generation behavior needed when a design team must generate an original illustration from a natural-language description. The large online retailer should instead use a supported image-generation model with a detailed permitted text prompt, followed by a check to submit a sample prompt and inspect generated image dimensions and relevance.

Incorrect Answers

A:

The option to assume all visual model endpoints accept arbitrary pixel sizes would normally serve a different aspect ratio control requirement, where a marketing app produces square assets when a wide banner is required. The alternative may be useful in its own context, but it does not directly supply assume a chat-only text endpoint automatically returns binary image assets for the stated decision.

C:

The alternative to use a supported image-generation model with a detailed permitted text prompt is a implementation step rather than the faulty assumption being diagnosed. This large online retailer failure-analysis alternative of use a supported image-generation model with a detailed permitted text prompt does not diagnose the particular text-driven image generation defect where a design team must generate an original illustration from a natural-language description; the keyed mistake is assume a chat-only text endpoint automatically returns binary image assets.

D:

The alternative to submit a sample prompt and inspect generated image dimensions and relevance is a targeted acceptance check step rather than the faulty assumption being diagnosed. This large online retailer failure-analysis alternative of submit a sample prompt and inspect generated image dimensions and relevance does not diagnose the particular text-driven image generation defect where a design team must generate an original illustration from a natural-language description; the keyed mistake is assume a chat-only text endpoint automatically returns binary image assets.

Question 653

The public library network is responding to a staged service rollout. A agent designer discovers this issue: A design team must generate an original illustration from a natural-language description. What corrective step should the team take to restore the intended functionality most directly?

  1. Select a compatible visual-generation deployment and retry with valid input
  2. Revise shots using the approved reference and controlled visual settings
  3. Submit a sample prompt and inspect generated image dimensions and relevance
  4. Assume a chat-only text endpoint automatically returns binary image assets

Explanation

Correct Answer

A:

The direct repair is to select a compatible visual-generation deployment and retry with valid input. Once the condition is corrected, the public library network should submit a sample prompt and inspect generated image dimensions and relevance. To reduce recurrence, it should also check permitted content and output asset type before rendering; those steps provide different recovery and prevention functions.

Incorrect Answers

B:

This approach relates to video reference consistency: a training video must show one product consistently across several generated shots. If the team chose it instead, the targeted check to submit a sample prompt and inspect generated image dimensions and relevance could still fail because the underlying required action is to select a compatible visual-generation deployment and retry with valid input.

C:

The action to submit a sample prompt and inspect generated image dimensions and relevance can contribute to text-driven image generation as a targeted acceptance check step; here, a design team must generate an original illustration from a natural-language description. During public library network remediation, the proposed step to submit a sample prompt and inspect generated image dimensions and relevance is not the direct restoration action for text-driven image generation when a design team must generate an original illustration from a natural-language description; the keyed repair is select a compatible visual-generation deployment and retry with valid input.

D:

During public library network remediation, the proposed step to assume a chat-only text endpoint automatically returns binary image assets is not the direct restoration action for text-driven image generation when a design team must generate an original illustration from a natural-language description; the keyed repair is select a compatible visual-generation deployment and retry with valid input. After remediation the team must submit a sample prompt and inspect generated image dimensions and relevance; the action in this option does not itself establish that target recovery outcome.

Question 654

A responsible-AI reviewer at a software consultancy wants the following issue to be prevented in later deployments: A design team must generate an original illustration from a natural-language description. Which enduring control or engineering guardrail BEST prevents recurrence?

  1. Assume a chat-only text endpoint automatically returns binary image assets
  2. Select a compatible visual-generation deployment and retry with valid input
  3. Check permitted content and output asset type before rendering
  4. Validate target dimensions and safe crop areas

Explanation

Correct Answer

C:

Standardizing the boundary to check permitted content and output asset type before rendering is preventive because the same text-driven image generation failure could otherwise return: a design team must generate an original illustration from a natural-language description. For this software consultancy, the service should also use a supported image-generation model with a detailed permitted text prompt, and tests should submit a sample prompt and inspect generated image dimensions and relevance so that the ongoing requirement is measurable.

Incorrect Answers

A:

For prevention at the software consultancy, the measure to assume a chat-only text endpoint automatically returns binary image assets is weaker or aimed at another issue than the text-driven image generation guardrail to check permitted content and output asset type before rendering needed when a design team must generate an original illustration from a natural-language description. The lasting guardrail for this specific threat is to check permitted content and output asset type before rendering, not simply to complete an adjacent service task.

B:

The action to select a compatible visual-generation deployment and retry with valid input can contribute to text-driven image generation as a corrective remediation step; here, a design team must generate an original illustration from a natural-language description. For prevention at the software consultancy, the measure to select a compatible visual-generation deployment and retry with valid input is weaker or aimed at another issue than the text-driven image generation guardrail to check permitted content and output asset type before rendering needed when a design team must generate an original illustration from a natural-language description.

D:

The action to validate target dimensions and safe crop areas addresses aspect ratio control rather than the decisive text-driven image generation condition. The software consultancy should not infer that fixing one adjacent AI service validates its unrelated workflow; the requested action is to check permitted content and output asset type before rendering.

Question 655

During a design meeting, a solution architect must accept a change for text-driven image generation. The issue to be addressed is: A design team must generate an original illustration from a natural-language description. Which concrete verification action demonstrates that the relevant behavior works?

  1. Submit a sample prompt and inspect generated image dimensions and relevance
  2. Use a supported image-generation model with a detailed permitted text prompt
  3. Assume a chat-only text endpoint automatically returns binary image assets
  4. Compare preceding and following frames and audio alignment

Explanation

Correct Answer

A:

The targeted acceptance test is to submit a sample prompt and inspect generated image dimensions and relevance. The design should first use a supported image-generation model with a detailed permitted text prompt. A robust regional hospital deployment also needs to check permitted content and output asset type before rendering; passing this check provides relevant evidence but does not certify every possible input or security risk.

Incorrect Answers

B:

The action to use a supported image-generation model with a detailed permitted text prompt can contribute to text-driven image generation as a implementation step; here, a design team must generate an original illustration from a natural-language description. In the regional hospital acceptance review, trying to use a supported image-generation model with a detailed permitted text prompt does not test the exact text-driven image generation condition where a design team must generate an original illustration from a natural-language description; the relevant test is to submit a sample prompt and inspect generated image dimensions and relevance.

C:

In the regional hospital acceptance review, trying to assume a chat-only text endpoint automatically returns binary image assets does not test the exact text-driven image generation condition where a design team must generate an original illustration from a natural-language description; the relevant test is to submit a sample prompt and inspect generated image dimensions and relevance. Acceptance should instead demonstrate that the service can submit a sample prompt and inspect generated image dimensions and relevance, using the described failure as the test criterion.

D:

In the regional hospital acceptance review, trying to compare preceding and following frames and audio alignment does not test the exact text-driven image generation condition where a design team must generate an original illustration from a natural-language description; the relevant test is to submit a sample prompt and inspect generated image dimensions and relevance. Without the actual change to submit a sample prompt and inspect generated image dimensions and relevance, the answer lacks the correct operational scope even if the alternate action is technically valid elsewhere.

Question 656

At a industrial maintenance team, the machine-learning developer is preparing a production incident. The concrete requirement is: A brand team wants new illustrations while retaining a supplied product silhouette. Which design action BEST addresses the stated need?

  1. Compare results with the input reference for intended visual features
  2. Use a supported reference-conditioned visual-generation workflow
  3. Promise exact preservation of geometry without using the reference input
  4. Use a supported image-generation model with a detailed permitted text prompt

Explanation

Correct Answer

B:

Choosing to use a supported reference-conditioned visual-generation workflow directly targets the described reference-image image generation gap: a brand team wants new illustrations while retaining a supplied product silhouette. As a lasting boundary, they also need to enforce rights for input images and avoid leaking reference media, because an implementation by itself does not prove safe operation.

Incorrect Answers

A:

The action to compare results with the input reference for intended visual features can contribute to reference-image image generation as a targeted acceptance check step; here, a brand team wants new illustrations while retaining a supplied product silhouette. Choosing to compare results with the input reference for intended visual features cannot directly replace the specified action to use a supported reference-conditioned visual-generation workflow when a brand team wants new illustrations while retaining a supplied product silhouette.

C:

Choosing to promise exact preservation of geometry without using the reference input cannot directly replace the specified action to use a supported reference-conditioned visual-generation workflow when a brand team wants new illustrations while retaining a supplied product silhouette. The actual implementation must use a supported reference-conditioned visual-generation workflow, because the stated service condition is that a brand team wants new illustrations while retaining a supplied product silhouette.

D:

Choosing to use a supported image-generation model with a detailed permitted text prompt would direct effort toward text-driven image generation, whose characteristic issue is that a design team must generate an original illustration from a natural-language description. The relevant corrective or preventive action here remains to use a supported reference-conditioned visual-generation workflow, with its own validation criteria.

Question 657

After a data protection inspection at a research university, the team observes this failure: A brand team wants new illustrations while retaining a supplied product silhouette. Which flawed assumption or implementation choice would MOST DIRECTLY explain why the intended outcome was not achieved?

  1. Use a supported reference-conditioned visual-generation workflow
  2. Compare results with the input reference for intended visual features
  3. Delete broad image areas when only a selected region needs correction
  4. Promise exact preservation of geometry without using the reference input

Explanation

Correct Answer

D:

The faulty assumption is promise exact preservation of geometry without using the reference input. A functioning alternative would be to use a supported reference-conditioned visual-generation workflow, and the relevant validation would be to compare results with the input reference for intended visual features.

Incorrect Answers

A:

The alternative to use a supported reference-conditioned visual-generation workflow is a implementation step rather than the faulty assumption being diagnosed. This research university failure-analysis alternative of use a supported reference-conditioned visual-generation workflow does not diagnose the particular reference-image image generation defect where a brand team wants new illustrations while retaining a supplied product silhouette; the keyed mistake is promise exact preservation of geometry without using the reference input.

B:

The alternative to compare results with the input reference for intended visual features is a targeted acceptance check step rather than the faulty assumption being diagnosed. This research university failure-analysis alternative of compare results with the input reference for intended visual features does not diagnose the particular reference-image image generation defect where a brand team wants new illustrations while retaining a supplied product silhouette; the keyed mistake is promise exact preservation of geometry without using the reference input.

C:

The option to delete broad image areas when only a selected region needs correction would normally serve a different object removal workflow requirement, where a product photo contains an unauthorized watermark-like artifact that must be removed under a valid license. The alternative may be useful in its own context, but it does not directly supply promise exact preservation of geometry without using the reference input for the stated decision.

Question 658

The education technology company is responding to a quality regression test. A quality assurance lead discovers this issue: A brand team wants new illustrations while retaining a supplied product silhouette. What corrective step should the team take to restore the intended functionality most directly?

  1. Supply the permitted reference asset and revise generation controls
  2. Call the appropriate visual editing path and inspect scene consistency
  3. Promise exact preservation of geometry without using the reference input
  4. Compare results with the input reference for intended visual features

Explanation

Correct Answer

A:

The appropriate remediation is to supply the permitted reference asset and revise generation controls, because the failure concerns reference-image image generation: a brand team wants new illustrations while retaining a supplied product silhouette. A meaningful verification should compare results with the input reference for intended visual features, and the durable operating rule should enforce rights for input images and avoid leaking reference media after the immediate fix.

Incorrect Answers

B:

This approach relates to prompt-driven image edit: a user requests a changed sky while retaining architecture and perspective. If the team chose it instead, the targeted check to compare results with the input reference for intended visual features could still fail because the underlying required action is to supply the permitted reference asset and revise generation controls.

C:

During education technology company remediation, the proposed step to promise exact preservation of geometry without using the reference input is not the direct restoration action for reference-image image generation when a brand team wants new illustrations while retaining a supplied product silhouette; the keyed repair is supply the permitted reference asset and revise generation controls. After remediation the team must compare results with the input reference for intended visual features; the action in this option does not itself establish that target recovery outcome.

D:

The action to compare results with the input reference for intended visual features can contribute to reference-image image generation as a targeted acceptance check step; here, a brand team wants new illustrations while retaining a supplied product silhouette. During education technology company remediation, the proposed step to compare results with the input reference for intended visual features is not the direct restoration action for reference-image image generation when a brand team wants new illustrations while retaining a supplied product silhouette; the keyed repair is supply the permitted reference asset and revise generation controls.

Question 659

A AI platform engineer at a pharmaceutical research group wants the following issue to be prevented in later deployments: A brand team wants new illustrations while retaining a supplied product silhouette. Which enduring control or engineering guardrail BEST prevents recurrence?

  1. Require timecoded edit scope and visual continuity review
  2. Supply the permitted reference asset and revise generation controls
  3. Enforce rights for input images and avoid leaking reference media
  4. Promise exact preservation of geometry without using the reference input

Explanation

Correct Answer

C:

The sustained prevention measure is to enforce rights for input images and avoid leaking reference media. The implementation can still use the separate action to use a supported reference-conditioned visual-generation workflow, but a guardrail constrains future changes. The pharmaceutical research group can measure whether the rule works by arranging to compare results with the input reference for intended visual features.

Incorrect Answers

A:

The action to require timecoded edit scope and visual continuity review addresses video edit segmentation rather than the decisive reference-image image generation condition. The pharmaceutical research group should not infer that fixing one adjacent AI service validates its unrelated workflow; the requested action is to enforce rights for input images and avoid leaking reference media.

B:

The action to supply the permitted reference asset and revise generation controls can contribute to reference-image image generation as a corrective remediation step; here, a brand team wants new illustrations while retaining a supplied product silhouette. For prevention at the pharmaceutical research group, the measure to supply the permitted reference asset and revise generation controls is weaker or aimed at another issue than the reference-image image generation guardrail to enforce rights for input images and avoid leaking reference media needed when a brand team wants new illustrations while retaining a supplied product silhouette.

D:

For prevention at the pharmaceutical research group, the measure to promise exact preservation of geometry without using the reference input is weaker or aimed at another issue than the reference-image image generation guardrail to enforce rights for input images and avoid leaking reference media needed when a brand team wants new illustrations while retaining a supplied product silhouette. The lasting guardrail for this specific threat is to enforce rights for input images and avoid leaking reference media, not simply to complete an adjacent service task.

Question 660

During a data protection inspection, a compliance specialist must accept a change for reference-image image generation. The issue to be addressed is: A brand team wants new illustrations while retaining a supplied product silhouette. Which concrete verification action demonstrates that the relevant behavior works?

  1. Submit a sample prompt and inspect generated image dimensions and relevance
  2. Enforce rights for input images and avoid leaking reference media
  3. Compare results with the input reference for intended visual features
  4. Use a supported reference-conditioned visual-generation workflow
  5. Promise exact preservation of geometry without using the reference input

Explanation

Correct Answer

C:

To verify this change, the travel support provider should compare results with the input reference for intended visual features. Operators still need the implementation to use a supported reference-conditioned visual-generation workflow and the preventive requirement to enforce rights for input images and avoid leaking reference media.

Incorrect Answers

A:

Choosing to submit a sample prompt and inspect generated image dimensions and relevance would direct effort toward text-driven image generation, whose characteristic issue is that a design team must generate an original illustration from a natural-language description. The relevant corrective or preventive action here remains to compare results with the input reference for intended visual features, with its own validation criteria.

B:

The action to enforce rights for input images and avoid leaking reference media can contribute to reference-image image generation as a ongoing preventive rule step; here, a brand team wants new illustrations while retaining a supplied product silhouette. In the travel support provider acceptance review, trying to enforce rights for input images and avoid leaking reference media does not test the exact reference-image image generation condition where a brand team wants new illustrations while retaining a supplied product silhouette; the relevant test is to compare results with the input reference for intended visual features.

D:

The action to use a supported reference-conditioned visual-generation workflow can contribute to reference-image image generation as a implementation step; here, a brand team wants new illustrations while retaining a supplied product silhouette. In the travel support provider acceptance review, trying to use a supported reference-conditioned visual-generation workflow does not test the exact reference-image image generation condition where a brand team wants new illustrations while retaining a supplied product silhouette; the relevant test is to compare results with the input reference for intended visual features.

E:

In the travel support provider acceptance review, trying to promise exact preservation of geometry without using the reference input does not test the exact reference-image image generation condition where a brand team wants new illustrations while retaining a supplied product silhouette; the relevant test is to compare results with the input reference for intended visual features. Acceptance should instead demonstrate that the service can compare results with the input reference for intended visual features, using the described failure as the test criterion.

Developing AI Apps and Agents on Azure (AI-103) Practice Tests