Microsoft AI-103 Dumps
Microsoft AI-103 Practice Test 58
Question 570
During a cost review, a responsible-AI reviewer must accept a change for retrieved context trimming. The issue to be addressed is: A RAG application sends too many irrelevant passages to the model. Which concrete verification action demonstrates that the relevant behavior works?
Explanation
Correct Answer
To verify this change, the software consultancy should measure answer groundedness with a smaller context budget. Operators still need the implementation to rerank and limit source chunks based on relevance and evidence coverage and the preventive requirement to monitor tokens and evidence recall for retrieval settings.
Incorrect Answers
For cost and quality dashboard, it can be sensible to run matched traffic samples and compare costs and ground-truth outcomes if operations cannot tell whether a model change saved money without hurting task success. Without the actual change to measure answer groundedness with a smaller context budget, the answer lacks the correct operational scope even if the alternate action is technically valid elsewhere.
In the software consultancy acceptance review, trying to send every candidate passage because a longer prompt always improves accuracy does not test the exact retrieved context trimming condition where a RAG application sends too many irrelevant passages to the model; the relevant test is to measure answer groundedness with a smaller context budget. Acceptance should instead demonstrate that the service can measure answer groundedness with a smaller context budget, using the described failure as the test criterion.
The action to rerank and limit source chunks based on relevance and evidence coverage can contribute to retrieved context trimming as a implementation step; here, a RAG application sends too many irrelevant passages to the model. In the software consultancy acceptance review, trying to rerank and limit source chunks based on relevance and evidence coverage does not test the exact retrieved context trimming condition where a RAG application sends too many irrelevant passages to the model; the relevant test is to measure answer groundedness with a smaller context budget.
Question 571
At a energy cooperative, the solution architect is preparing a support escalation. The concrete requirement is: An assistant repeats obsolete product prices from a cached answer. Which design action BEST addresses the stated need?
Explanation
Correct Answer
The proposed implementation is to include source version freshness in caching keys and expiration rules; this matches the operational requirement that an assistant repeats obsolete product prices from a cached answer. The team can update a product price and confirm the answer refreshes appropriately. The team should also document cache TTL and freshness requirements for dynamic topics.
Incorrect Answers
The action to update a product price and confirm the answer refreshes appropriately can contribute to semantic cache invalidation as a targeted acceptance check step; here, an assistant repeats obsolete product prices from a cached answer. Choosing to update a product price and confirm the answer refreshes appropriately cannot directly replace the specified action to include source version freshness in caching keys and expiration rules when an assistant repeats obsolete product prices from a cached answer.
Choosing to cache every response indefinitely without version awareness cannot directly replace the specified action to include source version freshness in caching keys and expiration rules when an assistant repeats obsolete product prices from a cached answer. The actual implementation must include source version freshness in caching keys and expiration rules, because the stated service condition is that an assistant repeats obsolete product prices from a cached answer.
Choosing to isolate test labels from the model generation context and audit evaluation inputs would direct effort toward evaluation data leakage, whose characteristic issue is that a model appears accurate because benchmark answers are embedded in prompts. The relevant corrective or preventive action here remains to include source version freshness in caching keys and expiration rules, with its own validation criteria.
Question 572
After a service audit at a industrial maintenance team, the team observes this failure: An assistant repeats obsolete product prices from a cached answer. Which flawed assumption or implementation choice would MOST DIRECTLY explain why the intended outcome was not achieved?
Explanation
Correct Answer
This design mistake is to cache every response indefinitely without version awareness; it conflicts with the actual semantic cache invalidation behavior needed when an assistant repeats obsolete product prices from a cached answer. The industrial maintenance team should instead include source version freshness in caching keys and expiration rules, followed by a check to update a product price and confirm the answer refreshes appropriately.
Incorrect Answers
The alternative to include source version freshness in caching keys and expiration rules is a implementation step rather than the faulty assumption being diagnosed. This industrial maintenance team failure-analysis alternative of include source version freshness in caching keys and expiration rules does not diagnose the particular semantic cache invalidation defect where an assistant repeats obsolete product prices from a cached answer; the keyed mistake is cache every response indefinitely without version awareness.
The alternative to document cache TTL and freshness requirements for dynamic topics is a ongoing preventive rule step rather than the faulty assumption being diagnosed. This industrial maintenance team failure-analysis alternative of document cache TTL and freshness requirements for dynamic topics does not diagnose the particular semantic cache invalidation defect where an assistant repeats obsolete product prices from a cached answer; the keyed mistake is cache every response indefinitely without version awareness.
The alternative to update a product price and confirm the answer refreshes appropriately is a targeted acceptance check step rather than the faulty assumption being diagnosed. This industrial maintenance team failure-analysis alternative of update a product price and confirm the answer refreshes appropriately does not diagnose the particular semantic cache invalidation defect where an assistant repeats obsolete product prices from a cached answer; the keyed mistake is cache every response indefinitely without version awareness.
This approach relates to token stream latency: users wait too long before seeing any response in a streaming chat app. If the team chose it instead, the targeted check to update a product price and confirm the answer refreshes appropriately could still fail because the underlying required action is to cache every response indefinitely without version awareness.
Question 573
The research university is responding to a security tabletop exercise. A AI application owner discovers this issue: An assistant repeats obsolete product prices from a cached answer. What corrective step should the team take to restore the intended functionality most directly?
Explanation
Correct Answer
The direct repair is to invalidate stale cached entries after the price source changes. Once the condition is corrected, the research university should update a product price and confirm the answer refreshes appropriately. To reduce recurrence, it should also document cache TTL and freshness requirements for dynamic topics; those steps provide different recovery and prevention functions.
Incorrect Answers
During research university remediation, the proposed step to restore the prior tested prompt and investigate failing examples is not the direct restoration action for semantic cache invalidation when an assistant repeats obsolete product prices from a cached answer; the keyed repair is invalidate stale cached entries after the price source changes. If the team chose it instead, the targeted check to update a product price and confirm the answer refreshes appropriately could still fail because the underlying required action is to invalidate stale cached entries after the price source changes.
During research university remediation, the proposed step to cache every response indefinitely without version awareness is not the direct restoration action for semantic cache invalidation when an assistant repeats obsolete product prices from a cached answer; the keyed repair is invalidate stale cached entries after the price source changes. After remediation the team must update a product price and confirm the answer refreshes appropriately; the action in this option does not itself establish that target recovery outcome.
The action to update a product price and confirm the answer refreshes appropriately can contribute to semantic cache invalidation as a targeted acceptance check step; here, an assistant repeats obsolete product prices from a cached answer. During research university remediation, the proposed step to update a product price and confirm the answer refreshes appropriately is not the direct restoration action for semantic cache invalidation when an assistant repeats obsolete product prices from a cached answer; the keyed repair is invalidate stale cached entries after the price source changes.
Question 574
A quality assurance lead at a education technology company wants the following issue to be prevented in later deployments: An assistant repeats obsolete product prices from a cached answer. Which enduring control or engineering guardrail BEST prevents recurrence?
Explanation
Correct Answer
Standardizing the boundary to document cache TTL and freshness requirements for dynamic topics is preventive because the same semantic cache invalidation failure could otherwise return: an assistant repeats obsolete product prices from a cached answer. For this education technology company, the service should also include source version freshness in caching keys and expiration rules, and tests should update a product price and confirm the answer refreshes appropriately so that the ongoing requirement is measurable.
Incorrect Answers
The action to limit parallel calls and test cancellation across branches addresses parallel tool retrieval rather than the decisive semantic cache invalidation condition. The education technology company should not infer that fixing one adjacent AI service validates its unrelated workflow; the requested action is to document cache TTL and freshness requirements for dynamic topics.
The action to invalidate stale cached entries after the price source changes can contribute to semantic cache invalidation as a corrective remediation step; here, an assistant repeats obsolete product prices from a cached answer. For prevention at the education technology company, the measure to invalidate stale cached entries after the price source changes is weaker or aimed at another issue than the semantic cache invalidation guardrail to document cache TTL and freshness requirements for dynamic topics needed when an assistant repeats obsolete product prices from a cached answer.
For prevention at the education technology company, the measure to cache every response indefinitely without version awareness is weaker or aimed at another issue than the semantic cache invalidation guardrail to document cache TTL and freshness requirements for dynamic topics needed when an assistant repeats obsolete product prices from a cached answer. The lasting guardrail for this specific threat is to document cache TTL and freshness requirements for dynamic topics, not simply to complete an adjacent service task.
Question 575
During a service audit, a AI platform engineer must accept a change for semantic cache invalidation. The issue to be addressed is: An assistant repeats obsolete product prices from a cached answer. Which concrete verification action demonstrates that the relevant behavior works?
Explanation
Correct Answer
The targeted acceptance test is to update a product price and confirm the answer refreshes appropriately. The design should first include source version freshness in caching keys and expiration rules. A robust pharmaceutical research group deployment also needs to document cache TTL and freshness requirements for dynamic topics; passing this check provides relevant evidence but does not certify every possible input or security risk.
Incorrect Answers
The action to include source version freshness in caching keys and expiration rules can contribute to semantic cache invalidation as a implementation step; here, an assistant repeats obsolete product prices from a cached answer. In the pharmaceutical research group acceptance review, trying to include source version freshness in caching keys and expiration rules does not test the exact semantic cache invalidation condition where an assistant repeats obsolete product prices from a cached answer; the relevant test is to update a product price and confirm the answer refreshes appropriately.
For guardrail parameter tuning, it can be sensible to compare accepted benign requests and blocked unsafe requests after tuning if an overly aggressive safety threshold rejects benign support questions. Without the actual change to update a product price and confirm the answer refreshes appropriately, the answer lacks the correct operational scope even if the alternate action is technically valid elsewhere.
In the pharmaceutical research group acceptance review, trying to cache every response indefinitely without version awareness does not test the exact semantic cache invalidation condition where an assistant repeats obsolete product prices from a cached answer; the relevant test is to update a product price and confirm the answer refreshes appropriately. Acceptance should instead demonstrate that the service can update a product price and confirm the answer refreshes appropriately, using the described failure as the test criterion.
Question 576
At a software consultancy, the compliance specialist is preparing a release acceptance review. The concrete requirement is: A new tool integration lets retrieved snippets influence unauthorized actions. Which design action BEST addresses the stated need?
Explanation
Correct Answer
Choosing to include indirect-injection adversarial cases in agent evaluation directly targets the described prompt injection regression gap: a new tool integration lets retrieved snippets influence unauthorized actions. As a lasting boundary, they also need to isolate tool output from instructions and enforce operation allowlists, because an implementation by itself does not prove safe operation.
Incorrect Answers
Choosing to version prompts and compare regression scores before production promotion would direct effort toward prompt version rollback, whose characteristic issue is that a revised system prompt changes correct responses on known tasks. The relevant corrective or preventive action here remains to include indirect-injection adversarial cases in agent evaluation, with its own validation criteria.
Choosing to evaluate only friendly queries that contain no adversarial tool content cannot directly replace the specified action to include indirect-injection adversarial cases in agent evaluation when a new tool integration lets retrieved snippets influence unauthorized actions. The actual implementation must include indirect-injection adversarial cases in agent evaluation, because the stated service condition is that a new tool integration lets retrieved snippets influence unauthorized actions.
The action to test a retrieved instruction attempting unauthorized export can contribute to prompt injection regression as a targeted acceptance check step; here, a new tool integration lets retrieved snippets influence unauthorized actions. Choosing to test a retrieved instruction attempting unauthorized export cannot directly replace the specified action to include indirect-injection adversarial cases in agent evaluation when a new tool integration lets retrieved snippets influence unauthorized actions.
Question 577
After a design meeting at a regional hospital, the team observes this failure: A new tool integration lets retrieved snippets influence unauthorized actions. Which flawed assumption or implementation choice would MOST DIRECTLY explain why the intended outcome was not achieved?
Explanation
Correct Answer
The faulty assumption is evaluate only friendly queries that contain no adversarial tool content. A functioning alternative would be to include indirect-injection adversarial cases in agent evaluation, and the relevant validation would be to test a retrieved instruction attempting unauthorized export.
Incorrect Answers
This regional hospital failure-analysis alternative of cache every response indefinitely without version awareness does not diagnose the particular prompt injection regression defect where a new tool integration lets retrieved snippets influence unauthorized actions; the keyed mistake is evaluate only friendly queries that contain no adversarial tool content. The alternative may be useful in its own context, but it does not directly supply evaluate only friendly queries that contain no adversarial tool content for the stated decision.
The alternative to test a retrieved instruction attempting unauthorized export is a targeted acceptance check step rather than the faulty assumption being diagnosed. This regional hospital failure-analysis alternative of test a retrieved instruction attempting unauthorized export does not diagnose the particular prompt injection regression defect where a new tool integration lets retrieved snippets influence unauthorized actions; the keyed mistake is evaluate only friendly queries that contain no adversarial tool content.
The alternative to include indirect-injection adversarial cases in agent evaluation is a implementation step rather than the faulty assumption being diagnosed. This regional hospital failure-analysis alternative of include indirect-injection adversarial cases in agent evaluation does not diagnose the particular prompt injection regression defect where a new tool integration lets retrieved snippets influence unauthorized actions; the keyed mistake is evaluate only friendly queries that contain no adversarial tool content.
Question 578
The insurance platform is responding to a staged service rollout. A data engineer discovers this issue: A new tool integration lets retrieved snippets influence unauthorized actions. What corrective step should the team take to restore the intended functionality most directly?
Explanation
Correct Answer
The appropriate remediation is to restore tool-boundary restrictions and replay adversarial cases, because the failure concerns prompt injection regression: a new tool integration lets retrieved snippets influence unauthorized actions. A meaningful verification should test a retrieved instruction attempting unauthorized export, and the durable operating rule should isolate tool output from instructions and enforce operation allowlists after the immediate fix.
Incorrect Answers
During insurance platform remediation, the proposed step to evaluate only friendly queries that contain no adversarial tool content is not the direct restoration action for prompt injection regression when a new tool integration lets retrieved snippets influence unauthorized actions; the keyed repair is restore tool-boundary restrictions and replay adversarial cases. After remediation the team must test a retrieved instruction attempting unauthorized export; the action in this option does not itself establish that target recovery outcome.
For retry backoff, it can be sensible to reduce concurrency and replay failed requests with idempotent retry behavior if a bursty application receives 429 responses from an inference endpoint. Without the actual change to restore tool-boundary restrictions and replay adversarial cases, the answer lacks the correct operational scope even if the alternate action is technically valid elsewhere.
The action to test a retrieved instruction attempting unauthorized export can contribute to prompt injection regression as a targeted acceptance check step; here, a new tool integration lets retrieved snippets influence unauthorized actions. During insurance platform remediation, the proposed step to test a retrieved instruction attempting unauthorized export is not the direct restoration action for prompt injection regression when a new tool integration lets retrieved snippets influence unauthorized actions; the keyed repair is restore tool-boundary restrictions and replay adversarial cases.
The action to isolate tool output from instructions and enforce operation allowlists can contribute to prompt injection regression as a ongoing preventive rule step; here, a new tool integration lets retrieved snippets influence unauthorized actions. During insurance platform remediation, the proposed step to isolate tool output from instructions and enforce operation allowlists is not the direct restoration action for prompt injection regression when a new tool integration lets retrieved snippets influence unauthorized actions; the keyed repair is restore tool-boundary restrictions and replay adversarial cases.
The action to include indirect-injection adversarial cases in agent evaluation can contribute to prompt injection regression as a implementation step; here, a new tool integration lets retrieved snippets influence unauthorized actions. During insurance platform remediation, the proposed step to include indirect-injection adversarial cases in agent evaluation is not the direct restoration action for prompt injection regression when a new tool integration lets retrieved snippets influence unauthorized actions; the keyed repair is restore tool-boundary restrictions and replay adversarial cases.
Question 579
A service administrator at a media platform wants the following issue to be prevented in later deployments: A new tool integration lets retrieved snippets influence unauthorized actions. Which enduring control or engineering guardrail BEST prevents recurrence?
Explanation
Correct Answer
The sustained prevention measure is to isolate tool output from instructions and enforce operation allowlists. The implementation can still use the separate action to include indirect-injection adversarial cases in agent evaluation, but a guardrail constrains future changes. The media platform can measure whether the rule works by arranging to test a retrieved instruction attempting unauthorized export.
Incorrect Answers
For prevention at the media platform, the measure to evaluate only friendly queries that contain no adversarial tool content is weaker or aimed at another issue than the prompt injection regression guardrail to isolate tool output from instructions and enforce operation allowlists needed when a new tool integration lets retrieved snippets influence unauthorized actions. The lasting guardrail for this specific threat is to isolate tool output from instructions and enforce operation allowlists, not simply to complete an adjacent service task.
The action to set trace retention and sensitive data minimization requirements addresses production trace correlation rather than the decisive prompt injection regression condition. The media platform should not infer that fixing one adjacent AI service validates its unrelated workflow; the requested action is to isolate tool output from instructions and enforce operation allowlists.
The action to restore tool-boundary restrictions and replay adversarial cases can contribute to prompt injection regression as a corrective remediation step; here, a new tool integration lets retrieved snippets influence unauthorized actions. For prevention at the media platform, the measure to restore tool-boundary restrictions and replay adversarial cases is weaker or aimed at another issue than the prompt injection regression guardrail to isolate tool output from instructions and enforce operation allowlists needed when a new tool integration lets retrieved snippets influence unauthorized actions.
Question 580
During a design meeting, a cloud developer must accept a change for prompt injection regression. The issue to be addressed is: A new tool integration lets retrieved snippets influence unauthorized actions. Which concrete verification action demonstrates that the relevant behavior works?
Explanation
Correct Answer
To verify this change, the customer support center should test a retrieved instruction attempting unauthorized export. Operators still need the implementation to include indirect-injection adversarial cases in agent evaluation and the preventive requirement to isolate tool output from instructions and enforce operation allowlists.
Incorrect Answers
The action to include indirect-injection adversarial cases in agent evaluation can contribute to prompt injection regression as a implementation step; here, a new tool integration lets retrieved snippets influence unauthorized actions. In the customer support center acceptance review, trying to include indirect-injection adversarial cases in agent evaluation does not test the exact prompt injection regression condition where a new tool integration lets retrieved snippets influence unauthorized actions; the relevant test is to test a retrieved instruction attempting unauthorized export.
In the customer support center acceptance review, trying to evaluate only friendly queries that contain no adversarial tool content does not test the exact prompt injection regression condition where a new tool integration lets retrieved snippets influence unauthorized actions; the relevant test is to test a retrieved instruction attempting unauthorized export. Acceptance should instead demonstrate that the service can test a retrieved instruction attempting unauthorized export, using the described failure as the test criterion.
For hallucination counterexample tests, it can be sensible to ask questions absent from the knowledge store and verify uncertainty if a new prompt handles known facts but invents missing ones. Without the actual change to test a retrieved instruction attempting unauthorized export, the answer lacks the correct operational scope even if the alternate action is technically valid elsewhere.
Developing AI Apps and Agents on Azure (AI-103) Practice Tests
- Microsoft AI-103 Practice Test 56 - 10 Questions (Last Reviewed 27-Sep-2026)
- Microsoft AI-103 Practice Test 57 - 10 Questions (Last Reviewed 22-Nov-2026)
- Microsoft AI-103 Practice Test 58 - 10 Questions (Last Reviewed 21-Nov-2026)
- Microsoft AI-103 Practice Test 59 - 10 Questions (Last Reviewed 20-Nov-2026)
- Microsoft AI-103 Practice Test 60 - 10 Questions (Last Reviewed 19-Nov-2026)