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 38

Question 370

During a source update, a cloud developer must accept a change for model tool response handling. The issue to be addressed is: A model emits a function call with missing required arguments. Which concrete verification action demonstrates that the relevant behavior works?

  1. Send malformed function arguments and confirm no side-effecting call occurs
  2. Validate tool-call JSON arguments against the declared schema before execution
  3. Ask questions absent from the corpus and confirm grounded uncertainty
  4. Execute incomplete tool calls and hope the API supplies safe defaults

Explanation

Correct Answer

A:

To verify this change, the logistics operator should send malformed function arguments and confirm no side-effecting call occurs. Operators still need the implementation to validate tool-call JSON arguments against the declared schema before execution and the preventive requirement to reject extra unauthorized fields and maintain versioned schemas.

Incorrect Answers

B:

The action to validate tool-call JSON arguments against the declared schema before execution can contribute to model tool response handling as a implementation step; here, a model emits a function call with missing required arguments. In the logistics operator acceptance review, trying to validate tool-call JSON arguments against the declared schema before execution does not test the exact model tool response handling condition where a model emits a function call with missing required arguments; the relevant test is to send malformed function arguments and confirm no side-effecting call occurs.

C:

For search score thresholds, it can be sensible to ask questions absent from the corpus and confirm grounded uncertainty if a model answers from low-relevance chunks despite no suitable document match. Without the actual change to send malformed function arguments and confirm no side-effecting call occurs, the answer lacks the correct operational scope even if the alternate action is technically valid elsewhere.

D:

In the logistics operator acceptance review, trying to execute incomplete tool calls and hope the API supplies safe defaults does not test the exact model tool response handling condition where a model emits a function call with missing required arguments; the relevant test is to send malformed function arguments and confirm no side-effecting call occurs. Acceptance should instead demonstrate that the service can send malformed function arguments and confirm no side-effecting call occurs, using the described failure as the test criterion.

Question 371

At a municipal services team, the site reliability engineer is preparing a security tabletop exercise. The concrete requirement is: A report summarizer invents conclusions not present in the supplied research report. Which design action BEST addresses the stated need?

  1. Permit plausible but unsupported claims to improve readability
  2. Constrain summary claims to evidence and explicitly indicate missing support
  3. Compare summary assertions to known supporting report passages
  4. Test supported sampling settings and prompt variations against quality and brand requirements

Explanation

Correct Answer

B:

The proposed implementation is to constrain summary claims to evidence and explicitly indicate missing support; this matches the operational requirement that a report summarizer invents conclusions not present in the supplied research report. The team can compare summary assertions to known supporting report passages. The team should also evaluate factual faithfulness rather than only summary fluency.

Incorrect Answers

A:

Choosing to permit plausible but unsupported claims to improve readability cannot directly replace the specified action to constrain summary claims to evidence and explicitly indicate missing support when a report summarizer invents conclusions not present in the supplied research report. The actual implementation must constrain summary claims to evidence and explicitly indicate missing support, because the stated service condition is that a report summarizer invents conclusions not present in the supplied research report.

C:

The action to compare summary assertions to known supporting report passages can contribute to grounded summarization as a targeted acceptance check step; here, a report summarizer invents conclusions not present in the supplied research report. Choosing to compare summary assertions to known supporting report passages cannot directly replace the specified action to constrain summary claims to evidence and explicitly indicate missing support when a report summarizer invents conclusions not present in the supplied research report.

D:

Choosing to test supported sampling settings and prompt variations against quality and brand requirements would direct effort toward sampling for creative drafting, whose characteristic issue is that a marketing draft assistant produces nearly identical copy despite requests for alternatives. The relevant corrective or preventive action here remains to constrain summary claims to evidence and explicitly indicate missing support, with its own validation criteria.

Question 372

A security engineer at a payment services team is conducting a support escalation concerning grounded summarization. The observed problem is: A report summarizer invents conclusions not present in the supplied research report. Which TWO actions together represent the needed implementation AND a durable preventive safeguard? (Select TWO answers.)

  1. Compare summary assertions to known supporting report passages
  2. Constrain summary claims to evidence and explicitly indicate missing support
  3. Evaluate factual faithfulness rather than only summary fluency
  4. Re-run summarization over retrieved report sections with claim evidence checks
  5. Apply supported semantic ranking on retrieved candidates with suitable text fields

Explanation

Correct Answers

B:

Choosing to constrain summary claims to evidence and explicitly indicate missing support directly targets the described grounded summarization gap: a report summarizer invents conclusions not present in the supplied research report. As a lasting boundary, they also need to evaluate factual faithfulness rather than only summary fluency, because an implementation by itself does not prove safe operation.

C:

Standardizing the boundary to evaluate factual faithfulness rather than only summary fluency is preventive because the same grounded summarization failure could otherwise return: a report summarizer invents conclusions not present in the supplied research report. For this payment services team, the service should also constrain summary claims to evidence and explicitly indicate missing support, and tests should compare summary assertions to known supporting report passages so that the ongoing requirement is measurable.

Incorrect Answers

A:

The alternative to compare summary assertions to known supporting report passages is a targeted acceptance check step rather than the faulty assumption being diagnosed. This payment services team failure-analysis alternative of compare summary assertions to known supporting report passages does not diagnose the particular grounded summarization defect where a report summarizer invents conclusions not present in the supplied research report; the keyed mistake is constrain summary claims to evidence and explicitly indicate missing support.

D:

The alternative to re-run summarization over retrieved report sections with claim evidence checks is a corrective remediation step rather than the faulty assumption being diagnosed. This payment services team failure-analysis alternative of re-run summarization over retrieved report sections with claim evidence checks does not diagnose the particular grounded summarization defect where a report summarizer invents conclusions not present in the supplied research report; the keyed mistake is constrain summary claims to evidence and explicitly indicate missing support.

E:

This approach relates to semantic reranking: search results contain relevant snippets far below unrelated vector matches. If the team chose it instead, the targeted check to compare summary assertions to known supporting report passages could still fail because the underlying required action is to constrain summary claims to evidence and explicitly indicate missing support.

Question 373

The transport authority is responding to a service audit. A agent designer discovers this issue: A report summarizer invents conclusions not present in the supplied research report. What corrective step should the team take to restore the intended functionality most directly?

  1. Permit plausible but unsupported claims to improve readability
  2. Compare summary assertions to known supporting report passages
  3. Reprocess the affected documents with better section-aware chunks
  4. Re-run summarization over retrieved report sections with claim evidence checks

Explanation

Correct Answer

D:

The direct repair is to re-run summarization over retrieved report sections with claim evidence checks. Once the condition is corrected, the transport authority should compare summary assertions to known supporting report passages. To reduce recurrence, it should also evaluate factual faithfulness rather than only summary fluency; those steps provide different recovery and prevention functions.

Incorrect Answers

A:

During transport authority remediation, the proposed step to permit plausible but unsupported claims to improve readability is not the direct restoration action for grounded summarization when a report summarizer invents conclusions not present in the supplied research report; the keyed repair is re-run summarization over retrieved report sections with claim evidence checks. After remediation the team must compare summary assertions to known supporting report passages; the action in this option does not itself establish that target recovery outcome.

B:

The action to compare summary assertions to known supporting report passages can contribute to grounded summarization as a targeted acceptance check step; here, a report summarizer invents conclusions not present in the supplied research report. During transport authority remediation, the proposed step to compare summary assertions to known supporting report passages is not the direct restoration action for grounded summarization when a report summarizer invents conclusions not present in the supplied research report; the keyed repair is re-run summarization over retrieved report sections with claim evidence checks.

C:

This approach relates to document chunk boundaries: a contract assistant misses a clause because paragraph boundaries split its context. If the team chose it instead, the targeted check to compare summary assertions to known supporting report passages could still fail because the underlying required action is to re-run summarization over retrieved report sections with claim evidence checks.

Question 374

A responsible-AI reviewer at a large online retailer wants the following issue to be prevented in later deployments: A report summarizer invents conclusions not present in the supplied research report. Which enduring control or engineering guardrail BEST prevents recurrence?

  1. Maintain dual query tests for identifier precision and paraphrase recall
  2. Compare summary assertions to known supporting report passages
  3. Permit plausible but unsupported claims to improve readability
  4. Constrain summary claims to evidence and explicitly indicate missing support
  5. Re-run summarization over retrieved report sections with claim evidence checks
  6. Evaluate factual faithfulness rather than only summary fluency

Explanation

Correct Answer

F:

Standardizing the boundary to evaluate factual faithfulness rather than only summary fluency is preventive because the same grounded summarization failure could otherwise return: a report summarizer invents conclusions not present in the supplied research report. For this large online retailer, the service should also constrain summary claims to evidence and explicitly indicate missing support, and tests should compare summary assertions to known supporting report passages so that the ongoing requirement is measurable.

Incorrect Answers

A:

Choosing to maintain dual query tests for identifier precision and paraphrase recall would direct effort toward hybrid retrieval for identifiers, whose characteristic issue is that users search for part numbers and semantically related troubleshooting symptoms. The relevant corrective or preventive action here remains to evaluate factual faithfulness rather than only summary fluency, with its own validation criteria.

B:

The action to compare summary assertions to known supporting report passages can contribute to grounded summarization as a targeted acceptance check step; here, a report summarizer invents conclusions not present in the supplied research report. For prevention at the large online retailer, the measure to compare summary assertions to known supporting report passages is weaker or aimed at another issue than the grounded summarization guardrail to evaluate factual faithfulness rather than only summary fluency needed when a report summarizer invents conclusions not present in the supplied research report.

C:

For prevention at the large online retailer, the measure to permit plausible but unsupported claims to improve readability is weaker or aimed at another issue than the grounded summarization guardrail to evaluate factual faithfulness rather than only summary fluency needed when a report summarizer invents conclusions not present in the supplied research report. The lasting guardrail for this specific threat is to evaluate factual faithfulness rather than only summary fluency, not simply to complete an adjacent service task.

D:

The action to constrain summary claims to evidence and explicitly indicate missing support can contribute to grounded summarization as a implementation step; here, a report summarizer invents conclusions not present in the supplied research report. For prevention at the large online retailer, the measure to constrain summary claims to evidence and explicitly indicate missing support is weaker or aimed at another issue than the grounded summarization guardrail to evaluate factual faithfulness rather than only summary fluency needed when a report summarizer invents conclusions not present in the supplied research report.

E:

The action to re-run summarization over retrieved report sections with claim evidence checks can contribute to grounded summarization as a corrective remediation step; here, a report summarizer invents conclusions not present in the supplied research report. For prevention at the large online retailer, the measure to re-run summarization over retrieved report sections with claim evidence checks is weaker or aimed at another issue than the grounded summarization guardrail to evaluate factual faithfulness rather than only summary fluency needed when a report summarizer invents conclusions not present in the supplied research report.

Question 375

During a support escalation, a solution architect must accept a change for grounded summarization. The issue to be addressed is: A report summarizer invents conclusions not present in the supplied research report. Which concrete verification action demonstrates that the relevant behavior works?

  1. Send a chat message through the deployed app and inspect the completion trace
  2. Constrain summary claims to evidence and explicitly indicate missing support
  3. Permit plausible but unsupported claims to improve readability
  4. Compare summary assertions to known supporting report passages

Explanation

Correct Answer

D:

The targeted acceptance test is to compare summary assertions to known supporting report passages. The design should first constrain summary claims to evidence and explicitly indicate missing support. A robust public library network deployment also needs to evaluate factual faithfulness rather than only summary fluency; passing this check provides relevant evidence but does not certify every possible input or security risk.

Incorrect Answers

A:

For model deployment for chat, it can be sensible to send a chat message through the deployed app and inspect the completion trace if a customer portal has no model endpoint capable of supporting conversational completion. Without the actual change to compare summary assertions to known supporting report passages, the answer lacks the correct operational scope even if the alternate action is technically valid elsewhere.

B:

The action to constrain summary claims to evidence and explicitly indicate missing support can contribute to grounded summarization as a implementation step; here, a report summarizer invents conclusions not present in the supplied research report. In the public library network acceptance review, trying to constrain summary claims to evidence and explicitly indicate missing support does not test the exact grounded summarization condition where a report summarizer invents conclusions not present in the supplied research report; the relevant test is to compare summary assertions to known supporting report passages.

C:

In the public library network acceptance review, trying to permit plausible but unsupported claims to improve readability does not test the exact grounded summarization condition where a report summarizer invents conclusions not present in the supplied research report; the relevant test is to compare summary assertions to known supporting report passages. Acceptance should instead demonstrate that the service can compare summary assertions to known supporting report passages, using the described failure as the test criterion.

Question 376

At a logistics operator, the machine-learning developer is preparing a staged service rollout. The concrete requirement is: A lengthy policy manual exceeds the model context window. Which design action BEST addresses the stated need?

  1. Ask about an appendix requirement and confirm it appears in the output
  2. Test supported sampling settings and prompt variations against quality and brand requirements
  3. Use a staged section-wise summarization workflow with cross-section reconciliation
  4. Truncate the end of the document and assume no key requirements are lost

Explanation

Correct Answer

C:

Choosing to use a staged section-wise summarization workflow with cross-section reconciliation directly targets the described long document summarization gap: a lengthy policy manual exceeds the model context window. As a lasting boundary, they also need to track source section coverage and avoid duplicate summaries, because an implementation by itself does not prove safe operation.

Incorrect Answers

A:

The action to ask about an appendix requirement and confirm it appears in the output can contribute to long document summarization as a targeted acceptance check step; here, a lengthy policy manual exceeds the model context window. Choosing to ask about an appendix requirement and confirm it appears in the output cannot directly replace the specified action to use a staged section-wise summarization workflow with cross-section reconciliation when a lengthy policy manual exceeds the model context window.

B:

Choosing to test supported sampling settings and prompt variations against quality and brand requirements would direct effort toward sampling for creative drafting, whose characteristic issue is that a marketing draft assistant produces nearly identical copy despite requests for alternatives. The relevant corrective or preventive action here remains to use a staged section-wise summarization workflow with cross-section reconciliation, with its own validation criteria.

D:

Choosing to truncate the end of the document and assume no key requirements are lost cannot directly replace the specified action to use a staged section-wise summarization workflow with cross-section reconciliation when a lengthy policy manual exceeds the model context window. The actual implementation must use a staged section-wise summarization workflow with cross-section reconciliation, because the stated service condition is that a lengthy policy manual exceeds the model context window.

Question 377

After a release acceptance review at a energy cooperative, the team observes this failure: A lengthy policy manual exceeds the model context window. Which flawed assumption or implementation choice would MOST DIRECTLY explain why the intended outcome was not achieved?

  1. Use a staged section-wise summarization workflow with cross-section reconciliation
  2. Truncate the end of the document and assume no key requirements are lost
  3. Ask about an appendix requirement and confirm it appears in the output
  4. Extract text alone and promise accurate chart interpretation without vision inputs

Explanation

Correct Answer

B:

The faulty assumption is truncate the end of the document and assume no key requirements are lost. A functioning alternative would be to use a staged section-wise summarization workflow with cross-section reconciliation, and the relevant validation would be to ask about an appendix requirement and confirm it appears in the output.

Incorrect Answers

A:

The alternative to use a staged section-wise summarization workflow with cross-section reconciliation is a implementation step rather than the faulty assumption being diagnosed. This energy cooperative failure-analysis alternative of use a staged section-wise summarization workflow with cross-section reconciliation does not diagnose the particular long document summarization defect where a lengthy policy manual exceeds the model context window; the keyed mistake is truncate the end of the document and assume no key requirements are lost.

C:

The alternative to ask about an appendix requirement and confirm it appears in the output is a targeted acceptance check step rather than the faulty assumption being diagnosed. This energy cooperative failure-analysis alternative of ask about an appendix requirement and confirm it appears in the output does not diagnose the particular long document summarization defect where a lengthy policy manual exceeds the model context window; the keyed mistake is truncate the end of the document and assume no key requirements are lost.

D:

This energy cooperative failure-analysis alternative of extract text alone and promise accurate chart interpretation without vision inputs does not diagnose the particular long document summarization defect where a lengthy policy manual exceeds the model context window; the keyed mistake is truncate the end of the document and assume no key requirements are lost. The alternative may be useful in its own context, but it does not directly supply truncate the end of the document and assume no key requirements are lost for the stated decision.

Question 378

The industrial maintenance team is responding to a design meeting. A quality assurance lead discovers this issue: A lengthy policy manual exceeds the model context window. What corrective step should the team take to restore the intended functionality most directly?

  1. Restore missing sections with bounded chunked processing
  2. Rebuild a contextual retrieval query from permitted prior turns
  3. Ask about an appendix requirement and confirm it appears in the output
  4. Truncate the end of the document and assume no key requirements are lost

Explanation

Correct Answer

A:

The appropriate remediation is to restore missing sections with bounded chunked processing, because the failure concerns long document summarization: a lengthy policy manual exceeds the model context window. A meaningful verification should ask about an appendix requirement and confirm it appears in the output, and the durable operating rule should track source section coverage and avoid duplicate summaries after the immediate fix.

Incorrect Answers

B:

This approach relates to multi-turn rag grounding: a user asks a follow-up whose meaning depends on prior context. If the team chose it instead, the targeted check to ask about an appendix requirement and confirm it appears in the output could still fail because the underlying required action is to restore missing sections with bounded chunked processing.

C:

The action to ask about an appendix requirement and confirm it appears in the output can contribute to long document summarization as a targeted acceptance check step; here, a lengthy policy manual exceeds the model context window. During industrial maintenance team remediation, the proposed step to ask about an appendix requirement and confirm it appears in the output is not the direct restoration action for long document summarization when a lengthy policy manual exceeds the model context window; the keyed repair is restore missing sections with bounded chunked processing.

D:

During industrial maintenance team remediation, the proposed step to truncate the end of the document and assume no key requirements are lost is not the direct restoration action for long document summarization when a lengthy policy manual exceeds the model context window; the keyed repair is restore missing sections with bounded chunked processing. After remediation the team must ask about an appendix requirement and confirm it appears in the output; the action in this option does not itself establish that target recovery outcome.

Question 379

A AI platform engineer at a research university wants the following issue to be prevented in later deployments: A lengthy policy manual exceeds the model context window. Which enduring control or engineering guardrail BEST prevents recurrence?

  1. Truncate the end of the document and assume no key requirements are lost
  2. Restore missing sections with bounded chunked processing
  3. Track source section coverage and avoid duplicate summaries
  4. Require tool timestamps and enforce access scope

Explanation

Correct Answer

C:

The sustained prevention measure is to track source section coverage and avoid duplicate summaries. The implementation can still use the separate action to use a staged section-wise summarization workflow with cross-section reconciliation, but a guardrail constrains future changes. The research university can measure whether the rule works by arranging to ask about an appendix requirement and confirm it appears in the output.

Incorrect Answers

A:

For prevention at the research university, the measure to truncate the end of the document and assume no key requirements are lost is weaker or aimed at another issue than the long document summarization guardrail to track source section coverage and avoid duplicate summaries needed when a lengthy policy manual exceeds the model context window. The lasting guardrail for this specific threat is to track source section coverage and avoid duplicate summaries, not simply to complete an adjacent service task.

B:

The action to restore missing sections with bounded chunked processing can contribute to long document summarization as a corrective remediation step; here, a lengthy policy manual exceeds the model context window. For prevention at the research university, the measure to restore missing sections with bounded chunked processing is weaker or aimed at another issue than the long document summarization guardrail to track source section coverage and avoid duplicate summaries needed when a lengthy policy manual exceeds the model context window.

D:

The action to require tool timestamps and enforce access scope addresses tool-augmented weather workflow rather than the decisive long document summarization condition. The research university should not infer that fixing one adjacent AI service validates its unrelated workflow; the requested action is to track source section coverage and avoid duplicate summaries.

Question 380

During a release acceptance review, a compliance specialist must accept a change for long document summarization. The issue to be addressed is: A lengthy policy manual exceeds the model context window. Which concrete verification action demonstrates that the relevant behavior works?

  1. Use a staged section-wise summarization workflow with cross-section reconciliation
  2. Ask for a known row-column intersection and compare the returned value
  3. Truncate the end of the document and assume no key requirements are lost
  4. Ask about an appendix requirement and confirm it appears in the output

Explanation

Correct Answer

D:

To verify this change, the education technology company should ask about an appendix requirement and confirm it appears in the output. Operators still need the implementation to use a staged section-wise summarization workflow with cross-section reconciliation and the preventive requirement to track source section coverage and avoid duplicate summaries.

Incorrect Answers

A:

The action to use a staged section-wise summarization workflow with cross-section reconciliation can contribute to long document summarization as a implementation step; here, a lengthy policy manual exceeds the model context window. In the education technology company acceptance review, trying to use a staged section-wise summarization workflow with cross-section reconciliation does not test the exact long document summarization condition where a lengthy policy manual exceeds the model context window; the relevant test is to ask about an appendix requirement and confirm it appears in the output.

B:

For grounded table question answering, it can be sensible to ask for a known row-column intersection and compare the returned value if a sales report assistant confuses figures across table headers. Without the actual change to ask about an appendix requirement and confirm it appears in the output, the answer lacks the correct operational scope even if the alternate action is technically valid elsewhere.

C:

In the education technology company acceptance review, trying to truncate the end of the document and assume no key requirements are lost does not test the exact long document summarization condition where a lengthy policy manual exceeds the model context window; the relevant test is to ask about an appendix requirement and confirm it appears in the output. Acceptance should instead demonstrate that the service can ask about an appendix requirement and confirm it appears in the output, using the described failure as the test criterion.

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