Microsoft AB-731 Practice Test 66
Question 650
At Riverbend Publishing, a customer service analyst needs data from an external ticket system. Evidence: The data is not necessarily available to Microsoft 365 Copilot without configuration. The business goal is to verify connector availability, authorization and freshness. The proposed solution must respect documented authority and organizational controls. Which approach most directly meets the stated business objective within the approved authority?
- Use a sanctioned connection or authorized export rather than inventing records
- Apply source and index security filtering before the LLM sees passages
- Design the workflow around entitled features or obtain approval for upgrades
- Draft in Outlook with verified language and require manual sending
Explanation
Correct Answer
A:
The data is not necessarily available to Microsoft 365 Copilot without configuration. Use a sanctioned connection or authorized export rather than inventing records.
Incorrect Answers
B:
“Apply source and index security filtering before the LLM sees passages” is a design or implementation response to the condition: Index query results lack caller-specific authorization filters. This question instead requires an action that can verify connector availability, authorization and freshness. The data is not necessarily available to Microsoft 365 Copilot without configuration. Use a sanctioned connection or authorized export rather than inventing records.
C:
“Design the workflow around entitled features or obtain approval for upgrades” is a design or implementation response to the condition: Feature entitlements vary by license, tenant and rollout. This question instead requires an action that can verify connector availability, authorization and freshness. The data is not necessarily available to Microsoft 365 Copilot without configuration. Use a sanctioned connection or authorized export rather than inventing records.
D:
“Draft in Outlook with verified language and require manual sending” tests a proposed change for the condition: A proposed discount is mentioned but not yet approved. This question instead requires an action that can verify connector availability, authorization and freshness. The data is not necessarily available to Microsoft 365 Copilot without configuration. Use a sanctioned connection or authorized export rather than inventing records.
Question 651
At Elmwood Banking, an employee forwards a vendor page containing instructions to ignore system policies. Evidence: The page is untrusted external content rather than a user-approved command. The business goal is to treat lower-trust input as data and defend against instruction injection. The operating team needs an ongoing check after the initial rollout. Which post-rollout signal is most important for detecting regression or governance drift?
- Monitor unusual source instructions and attempted tool misuse
- Track variance, exception rates and prompt drift
- Monitor false visual assertions and camera-domain drift
- Monitor occlusion failures and unseen packaging changes
Explanation
Correct Answer
A:
The page is untrusted external content rather than a user-approved command. Monitor unusual source instructions and attempted tool misuse.
Incorrect Answers
B:
“Track variance, exception rates and prompt drift” monitors this signal: Small phrasing changes cause inconsistent answers in a business-critical step. This question instead requires an action that can treat lower-trust input as data and defend against instruction injection. The page is untrusted external content rather than a user-approved command. Monitor unusual source instructions and attempted tool misuse.
C:
“Monitor false visual assertions and camera-domain drift” monitors this signal: Text descriptions alone omit visual indicators needed for triage. This question instead requires an action that can treat lower-trust input as data and defend against instruction injection. The page is untrusted external content rather than a user-approved command. Monitor unusual source instructions and attempted tool misuse.
D:
“Monitor occlusion failures and unseen packaging changes” monitors this signal: Visual interpretation matters more than long-form text creativity. This question instead requires an action that can treat lower-trust input as data and defend against instruction injection. The page is untrusted external content rather than a user-approved command. Monitor unusual source instructions and attempted tool misuse.
Question 652
At Orion Distribution, a research lead wants similar past cases retrieved before writing advice. Evidence: Similarity matching is distinct from composing a final recommendation. The business goal is to distinguish embeddings for retrieval from LLMs for generation. Only non-sensitive synthetic data and approved sandbox systems may be used. Which bounded trial with synthetic records best tests the proposed approach?
- Label fictional tickets including ambiguous mixed-urgency cases
- Load-test a simulated peak while monitoring error and safety rates
- Test semantically related fictional cases with different wording
- Compare retrieval on test questions before and after chunk changes
Explanation
Correct Answer
C:
Similarity matching is distinct from composing a final recommendation. Test semantically related fictional cases with different wording.
Incorrect Answers
A:
“Label fictional tickets including ambiguous mixed-urgency cases” tests a proposed change for the condition: The desired answer is a stable label rather than a free-form narrative. This question instead requires an action that can distinguish embeddings for retrieval from LLMs for generation. Similarity matching is distinct from composing a final recommendation. Test semantically related fictional cases with different wording.
B:
“Load-test a simulated peak while monitoring error and safety rates” tests a proposed change for the condition: The current manual process cannot expand rapidly without hiring. This question instead requires an action that can distinguish embeddings for retrieval from LLMs for generation. Similarity matching is distinct from composing a final recommendation. Test semantically related fictional cases with different wording.
D:
“Compare retrieval on test questions before and after chunk changes” tests a proposed change for the condition: Large irrelevant text consumes context and obscures the answer. This question instead requires an action that can distinguish embeddings for retrieval from LLMs for generation. Similarity matching is distinct from composing a final recommendation. Test semantically related fictional cases with different wording.
Question 653
At Silverpine Group, a proposed knowledge assistant sends an entire library to the model each time. Evidence: Large retrieved context increases token spending and response time. The business goal is to budget retrieval and generation costs together. Only non-sensitive synthetic data and approved sandbox systems may be used. Which bounded trial with synthetic records best tests the proposed approach?
- Compare full-library prompting with top-k retrieval on mock files
- Compare top-three and top-ten retrieval on synthetic questions
- Run a synthetic migration with renamed metadata fields
- Cluster fictional feedback including one emerging novel issue
Explanation
Correct Answer
A:
Large retrieved context increases token spending and response time. Compare full-library prompting with top-k retrieval on mock files.
Incorrect Answers
B:
“Compare top-three and top-ten retrieval on synthetic questions” tests a proposed change for the condition: Retrieving many passages improves coverage while increasing noise. This question instead requires an action that can budget retrieval and generation costs together. Large retrieved context increases token spending and response time. Compare full-library prompting with top-k retrieval on mock files.
C:
“Run a synthetic migration with renamed metadata fields” tests a proposed change for the condition: Retrieved field meanings changed even though model version is unchanged. This question instead requires an action that can budget retrieval and generation costs together. Large retrieved context increases token spending and response time. Compare full-library prompting with top-k retrieval on mock files.
D:
“Cluster fictional feedback including one emerging novel issue” tests a proposed change for the condition: There are no predefined labels and the first goal is discovering patterns. This question instead requires an action that can budget retrieval and generation costs together. Large retrieved context increases token spending and response time. Compare full-library prompting with top-k retrieval on mock files.
Question 654
At Cascade Utilities, a policy director needs a comparative industry briefing from many sources. Evidence: The question requires multi-step evidence gathering rather than quick lookup. The business goal is to select Researcher for deeper cited multi-source synthesis where available. The sponsor needs an evidence-based diagnosis rather than a guessed fix. Which initial investigation best identifies the controlling constraint before any solution is chosen?
- Identify protected image data and processing agreements
- List which facts need public search versus work file access
- Map required data, actions, users and existing agents
- Count required sources and assess breadth, recency and permissions
Explanation
Correct Answer
D:
The question requires multi-step evidence gathering rather than quick lookup. Count required sources and assess breadth, recency and permissions.
Incorrect Answers
A:
“Identify protected image data and processing agreements” tests a proposed change for the condition: A less sensitive evaluation could answer the technical question. This question instead requires an action that can select Researcher for deeper cited multi-source synthesis where available. The question requires multi-step evidence gathering rather than quick lookup. Count required sources and assess breadth, recency and permissions.
B:
“List which facts need public search versus work file access” is a diagnostic step for the reported condition: Public web claims and workplace records have different provenance. This question instead requires an action that can select Researcher for deeper cited multi-source synthesis where available. The question requires multi-step evidence gathering rather than quick lookup. Count required sources and assess breadth, recency and permissions.
C:
“Map required data, actions, users and existing agents” tests a proposed change for the condition: Base Copilot chat cannot directly implement every business-specific action. This question instead requires an action that can select Researcher for deeper cited multi-source synthesis where available. The question requires multi-step evidence gathering rather than quick lookup. Count required sources and assess breadth, recency and permissions.
Question 655
At Lakeshore Commerce, a procurement team gets different classifications for the same standardized request. Evidence: Small phrasing changes cause inconsistent answers in a business-critical step. The business goal is to test consistency rather than trusting one convincing example. The sponsor needs an evidence-based diagnosis rather than a guessed fix. Which initial investigation best identifies the controlling constraint before any solution is chosen?
- Record baseline cost per resolved case, volume and quality
- Identify prompts that should have equivalent outcomes
- Measure relevant hits and irrelevant context on labeled queries
- Distinguish exact extraction from creative brainstorming needs
Explanation
Correct Answer
B:
Small phrasing changes cause inconsistent answers in a business-critical step. Identify prompts that should have equivalent outcomes.
Incorrect Answers
A:
“Record baseline cost per resolved case, volume and quality” is a diagnostic step for the reported condition: No baseline handling time, error rate or deployment cost is recorded. This question instead requires an action that can test consistency rather than trusting one convincing example. Small phrasing changes cause inconsistent answers in a business-critical step. Identify prompts that should have equivalent outcomes.
C:
“Measure relevant hits and irrelevant context on labeled queries” is a diagnostic step for the reported condition: Retrieving many passages improves coverage while increasing noise. This question instead requires an action that can test consistency rather than trusting one convincing example. Small phrasing changes cause inconsistent answers in a business-critical step. Identify prompts that should have equivalent outcomes.
D:
“Distinguish exact extraction from creative brainstorming needs” is a diagnostic step for the reported condition: Creative variation adds no value and can introduce mismatches. This question instead requires an action that can test consistency rather than trusting one convincing example. Small phrasing changes cause inconsistent answers in a business-critical step. Identify prompts that should have equivalent outcomes.
Question 656
At Lakeshore Commerce, a department deploys an agent that can approve purchase orders. Evidence: The agent's action scope could exceed its intended advisory role. The business goal is to govern tool use and high-impact autonomy explicitly. The proposed solution must respect documented authority and organizational controls. Which approach most directly meets the stated business objective within the approved authority?
- Set budgets and review usage through supported cost management
- Require impact assessment, limits, notice and approval
- Reduce data capture and restrict retention to approved purpose
- Restrict approval actions behind human authorization and audit
Explanation
Correct Answer
D:
The agent's action scope could exceed its intended advisory role. Restrict approval actions behind human authorization and audit. An advisory assistant and an autonomous agent have different failure impact. Restrict write tools, approval thresholds and action logs in line with the consequences of each operation.
Incorrect Answers
A:
“Set budgets and review usage through supported cost management” is a design or implementation response to the condition: Consumption can rise faster than the annual forecast. This question instead requires an action that can govern tool use and high-impact autonomy explicitly. The agent's action scope could exceed its intended advisory role. Restrict approval actions behind human authorization and audit. An advisory assistant and an autonomous agent have different failure impact. Restrict write tools, approval thresholds and action logs in line with the consequences of each operation.
B:
“Require impact assessment, limits, notice and approval” is a design or implementation response to the condition: The proposed system analyzes communications and behavior at scale. This question instead requires an action that can govern tool use and high-impact autonomy explicitly. The agent's action scope could exceed its intended advisory role. Restrict approval actions behind human authorization and audit. An advisory assistant and an autonomous agent have different failure impact. Restrict write tools, approval thresholds and action logs in line with the consequences of each operation.
C:
“Reduce data capture and restrict retention to approved purpose” is a design or implementation response to the condition: The project only needs agreed project decisions. This question instead requires an action that can govern tool use and high-impact autonomy explicitly. The agent's action scope could exceed its intended advisory role. Restrict approval actions behind human authorization and audit. An advisory assistant and an autonomous agent have different failure impact. Restrict write tools, approval thresholds and action logs in line with the consequences of each operation.
Question 657
At Blue Ridge Technology, a hiring assistant pilot uses historical examples from one office only. Evidence: The target deployment covers regions with very different job families. The business goal is to check evaluation representativeness before organization-wide claims. The operating team needs an ongoing check after the initial rollout. Which post-rollout signal is most important for detecting regression or governance drift?
- Monitor AI suggestions that conflict with accounting controls
- Monitor data distribution drift and uncovered roles
- Monitor unintended transcript sharing and retention exceptions
- Track precision for paraphrased and exact-match queries
Explanation
Correct Answer
B:
The target deployment covers regions with very different job families. Monitor data distribution drift and uncovered roles. Accuracy measured on one region or job family may mask poor performance elsewhere. Sample intended deployment cohorts and report subgroup failures before claiming enterprise representativeness.
Incorrect Answers
A:
“Monitor AI suggestions that conflict with accounting controls” monitors this signal: The reconciliation is deterministic arithmetic with known control rules. This question instead requires an action that can check evaluation representativeness before organization-wide claims. The target deployment covers regions with very different job families. Monitor data distribution drift and uncovered roles. Accuracy measured on one region or job family may mask poor performance elsewhere. Sample intended deployment cohorts and report subgroup failures before claiming enterprise representativeness.
C:
“Monitor unintended transcript sharing and retention exceptions” monitors this signal: Recording contains names and confidential discussions. This question instead requires an action that can check evaluation representativeness before organization-wide claims. The target deployment covers regions with very different job families. Monitor data distribution drift and uncovered roles. Accuracy measured on one region or job family may mask poor performance elsewhere. Sample intended deployment cohorts and report subgroup failures before claiming enterprise representativeness.
D:
“Track precision for paraphrased and exact-match queries” monitors this signal: Matching literal keywords alone fails for synonymous wording. This question instead requires an action that can check evaluation representativeness before organization-wide claims. The target deployment covers regions with very different job families. Monitor data distribution drift and uncovered roles. Accuracy measured on one region or job family may mask poor performance elsewhere. Sample intended deployment cohorts and report subgroup failures before claiming enterprise representativeness.
Question 658
At Summit Manufacturing, a marketing lead assumes public Copilot can read a protected SharePoint strategy. Evidence: The service has not been authorized to access the requested private file. The business goal is to distinguish public web answers from permissioned work grounding. Only non-sensitive synthetic data and approved sandbox systems may be used. Which bounded trial with synthetic records best tests the proposed approach?
- Test recurring queries against a fictional notebook source set
- Trial with public articles and fictional internal project notes
- Test with a synthetic private file visible only to one user
- Search fake regulations using a code and a synonym
Explanation
Correct Answer
C:
The service has not been authorized to access the requested private file. Test with a synthetic private file visible only to one user.
Incorrect Answers
A:
“Test recurring queries against a fictional notebook source set” tests a proposed change for the condition: Unorganized ad hoc chats repeatedly lose contextual references. This question instead requires an action that can distinguish public web answers from permissioned work grounding. The service has not been authorized to access the requested private file. Test with a synthetic private file visible only to one user.
B:
“Trial with public articles and fictional internal project notes” tests a proposed change for the condition: Public web claims and workplace records have different provenance. This question instead requires an action that can distinguish public web answers from permissioned work grounding. The service has not been authorized to access the requested private file. Test with a synthetic private file visible only to one user.
D:
“Search fake regulations using a code and a synonym” tests a proposed change for the condition: Vector similarity alone is not enough to preserve identifiers. This question instead requires an action that can distinguish public web answers from permissioned work grounding. The service has not been authorized to access the requested private file. Test with a synthetic private file visible only to one user.
Question 659
At Evergreen Research, a language model begins answering poorly after a source system changes schema. Evidence: Retrieved field meanings changed even though model version is unchanged. The business goal is to diagnose data pipeline changes before blaming model weights. The operating team needs an ongoing check after the initial rollout. Which post-rollout signal is most important for detecting regression or governance drift?
- Monitor unintended transcript sharing and retention exceptions
- Monitor leaked tokens and abnormal inference charges
- Monitor schema changes and retrieval quality by source
- Monitor transcription errors, consent settings and missing audio segments
Explanation
Correct Answer
C:
Retrieved field meanings changed even though model version is unchanged. Monitor schema changes and retrieval quality by source. Changing an upstream field definition changes the data the model sees even if the model weights are fixed. Check mapping lineage, schema and retrieval output before attributing the regression to model behavior.
Incorrect Answers
A:
“Monitor unintended transcript sharing and retention exceptions” monitors this signal: Recording contains names and confidential discussions. This question instead requires an action that can diagnose data pipeline changes before blaming model weights. Retrieved field meanings changed even though model version is unchanged. Monitor schema changes and retrieval quality by source. Changing an upstream field definition changes the data the model sees even if the model weights are fixed. Check mapping lineage, schema and retrieval output before attributing the regression to model behavior.
B:
“Monitor leaked tokens and abnormal inference charges” monitors this signal: The credentials enable unauthorized paid inference requests. This question instead requires an action that can diagnose data pipeline changes before blaming model weights. Retrieved field meanings changed even though model version is unchanged. Monitor schema changes and retrieval quality by source. Changing an upstream field definition changes the data the model sees even if the model weights are fixed. Check mapping lineage, schema and retrieval output before attributing the regression to model behavior.
D:
“Monitor transcription errors, consent settings and missing audio segments” monitors this signal: Poor transcription can propagate incorrect speaker attribution into summaries. This question instead requires an action that can diagnose data pipeline changes before blaming model weights. Retrieved field meanings changed even though model version is unchanged. Monitor schema changes and retrieval quality by source. Changing an upstream field definition changes the data the model sees even if the model weights are fixed. Check mapping lineage, schema and retrieval output before attributing the regression to model behavior.
Question 660
At BrightPath Media, a bank purchases an AI service without reviewing vendor processing terms. Evidence: Contract data and prompts may be handled under unclear commitments. The business goal is to review third-party AI risks before procurement approval. The operating team needs an ongoing check after the initial rollout. Which post-rollout signal is most important for detecting regression or governance drift?
- Monitor vendor incident notices and material terms changes
- Monitor usage spikes and billing reconciliation delays
- Monitor unresolved complaints and disparate correction times
- Monitor rubber-stamp review and bypassed approvals
Explanation
Correct Answer
A:
Contract data and prompts may be handled under unclear commitments. Monitor vendor incident notices and material terms changes.
Incorrect Answers
B:
“Monitor usage spikes and billing reconciliation delays” monitors this signal: Consumption can rise faster than the annual forecast. This question instead requires an action that can review third-party AI risks before procurement approval. Contract data and prompts may be handled under unclear commitments. Monitor vendor incident notices and material terms changes.
C:
“Monitor unresolved complaints and disparate correction times” monitors this signal: Users may be unable to correct harmful mistakes. This question instead requires an action that can review third-party AI risks before procurement approval. Contract data and prompts may be handled under unclear commitments. Monitor vendor incident notices and material terms changes.
D:
“Monitor rubber-stamp review and bypassed approvals” monitors this signal: Ambiguous policy exceptions can harm individuals. This question instead requires an action that can review third-party AI risks before procurement approval. Contract data and prompts may be handled under unclear commitments. Monitor vendor incident notices and material terms changes.