Microsoft AB-731 Practice Test 44
Question 430
At Redwood Consulting, a legal department searches by exact contract number and semantically similar clauses. Evidence: Vector-only matching risks missing literal identifiers. The business goal is to consider hybrid search combining keyword and semantic retrieval. The project owner requires a measurable acceptance criterion before signoff. Which evidence would most convincingly demonstrate that the requirement has been met?
- The public-only experience never claims unseen private details
- Contributors cannot publish without approval while tests remain possible
- Risky outputs are handled safely without rejecting ordinary requests
- Both exact and semantic requirements retrieve correct approved text
Explanation
Correct Answer
D:
Vector-only matching risks missing literal identifiers. Both exact and semantic requirements retrieve correct approved text. Vector similarity can find concepts while missing exact part numbers or codes. Combining lexical and semantic retrieval retains identifier precision and can improve relevant passage recall.
Incorrect Answers
A:
“The public-only experience never claims unseen private details” validates an outcome in a case where the evidence is: The service has not been authorized to access the requested private file. This question instead requires an action that can consider hybrid search combining keyword and semantic retrieval. Vector-only matching risks missing literal identifiers. Both exact and semantic requirements retrieve correct approved text. Vector similarity can find concepts while missing exact part numbers or codes. Combining lexical and semantic retrieval retains identifier precision and can improve relevant passage recall.
B:
“Contributors cannot publish without approval while tests remain possible” validates an outcome in a case where the evidence is: Excess access can lead to unreviewed deployments and data exposure. This question instead requires an action that can consider hybrid search combining keyword and semantic retrieval. Vector-only matching risks missing literal identifiers. Both exact and semantic requirements retrieve correct approved text. Vector similarity can find concepts while missing exact part numbers or codes. Combining lexical and semantic retrieval retains identifier precision and can improve relevant passage recall.
C:
“Risky outputs are handled safely without rejecting ordinary requests” validates an outcome in a case where the evidence is: The release needs safety checks beyond normal accuracy tests. This question instead requires an action that can consider hybrid search combining keyword and semantic retrieval. Vector-only matching risks missing literal identifiers. Both exact and semantic requirements retrieve correct approved text. Vector similarity can find concepts while missing exact part numbers or codes. Combining lexical and semantic retrieval retains identifier precision and can improve relevant passage recall.
Question 431
At Bridgewater Logistics, a generated compliance answer cites an unrelated section of the handbook. Evidence: Retrieval produced a plausible paragraph but not supporting evidence. The business goal is to require claim-level grounding and source verification. The project owner requires a measurable acceptance criterion before signoff. Which evidence would most convincingly demonstrate that the requirement has been met?
- A clean holdout provides a credible lower performance estimate
- Chosen approach meets quality targets at lower cost and acceptable latency
- Release is blocked if safety thresholds fail despite accurate answers
- Answers identify the right clause or explicitly express uncertainty
Explanation
Correct Answer
D:
Retrieval produced a plausible paragraph but not supporting evidence. Answers identify the right clause or explicitly express uncertainty.
Incorrect Answers
A:
“A clean holdout provides a credible lower performance estimate” validates an outcome in a case where the evidence is: The same customer records appear in tuning and testing data. This question instead requires an action that can require claim-level grounding and source verification. Retrieval produced a plausible paragraph but not supporting evidence. Answers identify the right clause or explicitly express uncertainty.
B:
“Chosen approach meets quality targets at lower cost and acceptable latency” validates an outcome in a case where the evidence is: Simple tags may be handled by a lower-cost capable model. This question instead requires an action that can require claim-level grounding and source verification. Retrieval produced a plausible paragraph but not supporting evidence. Answers identify the right clause or explicitly express uncertainty.
C:
“Release is blocked if safety thresholds fail despite accurate answers” validates an outcome in a case where the evidence is: Data protection failures are not reflected in the accuracy metric. This question instead requires an action that can require claim-level grounding and source verification. Retrieval produced a plausible paragraph but not supporting evidence. Answers identify the right clause or explicitly express uncertainty.
Question 432
At Tamarack Logistics, an organization imposes the same lengthy review on all AI use cases. Evidence: A low-risk note formatter and high-stakes hiring scorer carry different risks. The business goal is to scale governance depth to documented impact and autonomy. The project owner requires a measurable acceptance criterion before signoff. Which evidence would most convincingly demonstrate that the requirement has been met?
- Similar qualified applicants receive comparable screening treatment
- High-impact proposals receive appropriately deeper scrutiny
- Costs align with necessary capabilities and avoid redundant seats
- Users can complete core tasks without inaccessible controls
Explanation
Correct Answer
B:
A low-risk note formatter and high-stakes hiring scorer carry different risks. High-impact proposals receive appropriately deeper scrutiny.
Incorrect Answers
A:
“Similar qualified applicants receive comparable screening treatment” validates an outcome in a case where the evidence is: Performance differs sharply across legally relevant applicant groups. This question instead requires an action that can scale governance depth to documented impact and autonomy. A low-risk note formatter and high-stakes hiring scorer carry different risks. High-impact proposals receive appropriately deeper scrutiny.
C:
“Costs align with necessary capabilities and avoid redundant seats” validates an outcome in a case where the evidence is: Usage patterns vary and some workflows may use included chat capabilities. This question instead requires an action that can scale governance depth to documented impact and autonomy. A low-risk note formatter and high-stakes hiring scorer carry different risks. High-impact proposals receive appropriately deeper scrutiny.
D:
“Users can complete core tasks without inaccessible controls” validates an outcome in a case where the evidence is: Keyboard navigation and alternative text were not tested. This question instead requires an action that can scale governance depth to documented impact and autonomy. A low-risk note formatter and high-stakes hiring scorer carry different risks. High-impact proposals receive appropriately deeper scrutiny.
Question 433
At Solstice Retail, a finance director sees model spending rise despite stable request counts. Evidence: Longer retrieved passages and responses increase processed input and output tokens. The business goal is to estimate consumption using token counts and actual price meters. Only non-sensitive synthetic data and approved sandbox systems may be used. Which bounded trial with synthetic records best tests the proposed approach?
- Insert a deliberately unsupported statistic into a fictional draft and review it
- Test a harmless fake exfiltration instruction in a sandbox document
- Compare an extracted clause with an AI paraphrase of the same mock contract
- Compare costs of two equivalent prompts with different context lengths
Explanation
Correct Answer
D:
Longer retrieved passages and responses increase processed input and output tokens. Compare costs of two equivalent prompts with different context lengths. Retrieval length and generated response length both contribute billable tokens. Estimate workload cost from measured input/output token distributions and the actual deployed model tariff, not a single average prompt.
Incorrect Answers
A:
“Insert a deliberately unsupported statistic into a fictional draft and review it” tests a proposed change for the condition: The cited source does not contain the specific numerical assertion. This question instead requires an action that can estimate consumption using token counts and actual price meters. Longer retrieved passages and responses increase processed input and output tokens. Compare costs of two equivalent prompts with different context lengths. Retrieval length and generated response length both contribute billable tokens. Estimate workload cost from measured input/output token distributions and the actual deployed model tariff, not a single average prompt.
B:
“Test a harmless fake exfiltration instruction in a sandbox document” tests a proposed change for the condition: The embedded directive came from lower-trust document content. This question instead requires an action that can estimate consumption using token counts and actual price meters. Longer retrieved passages and responses increase processed input and output tokens. Compare costs of two equivalent prompts with different context lengths. Retrieval length and generated response length both contribute billable tokens. Estimate workload cost from measured input/output token distributions and the actual deployed model tariff, not a single average prompt.
C:
“Compare an extracted clause with an AI paraphrase of the same mock contract” tests a proposed change for the condition: Paraphrasing would change obligations and cannot replace the signed source. This question instead requires an action that can estimate consumption using token counts and actual price meters. Longer retrieved passages and responses increase processed input and output tokens. Compare costs of two equivalent prompts with different context lengths. Retrieval length and generated response length both contribute billable tokens. Estimate workload cost from measured input/output token distributions and the actual deployed model tariff, not a single average prompt.
Question 434
At Cascade Utilities, a team stores model API keys in shared spreadsheets. Evidence: The credentials enable unauthorized paid inference requests. The business goal is to use credential vaulting and least privilege rather than shared secrets. 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?
- Open cited passages and compare each claim to supporting text
- Identify exposed secrets and account permissions
- Measure relevant hits and irrelevant context on labeled queries
- Inventory supported languages and representative customer language mixes
Explanation
Correct Answer
B:
The credentials enable unauthorized paid inference requests. Identify exposed secrets and account permissions. A leaked API credential can permit unauthorized requests and usage charges until revoked. Store secrets in managed credentials, scope access narrowly, rotate exposed keys and monitor anomalous consumption.
Incorrect Answers
A:
“Open cited passages and compare each claim to supporting text” is a diagnostic step for the reported condition: Retrieval produced a plausible paragraph but not supporting evidence. This question instead requires an action that can use credential vaulting and least privilege rather than shared secrets. The credentials enable unauthorized paid inference requests. Identify exposed secrets and account permissions. A leaked API credential can permit unauthorized requests and usage charges until revoked. Store secrets in managed credentials, scope access narrowly, rotate exposed keys and monitor anomalous consumption.
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 use credential vaulting and least privilege rather than shared secrets. The credentials enable unauthorized paid inference requests. Identify exposed secrets and account permissions. A leaked API credential can permit unauthorized requests and usage charges until revoked. Store secrets in managed credentials, scope access narrowly, rotate exposed keys and monitor anomalous consumption.
D:
“Inventory supported languages and representative customer language mixes” is a diagnostic step for the reported condition: Evaluation data covers only English and may hide translation losses. This question instead requires an action that can use credential vaulting and least privilege rather than shared secrets. The credentials enable unauthorized paid inference requests. Identify exposed secrets and account permissions. A leaked API credential can permit unauthorized requests and usage charges until revoked. Store secrets in managed credentials, scope access narrowly, rotate exposed keys and monitor anomalous consumption.
Question 435
At Prairie Health, a department refuses an AI assistant after one fabricated answer. Evidence: Users have no transparent examples of error correction or escalation. The business goal is to rebuild trust through evidence, safeguards and iterative feedback. Only non-sensitive synthetic data and approved sandbox systems may be used. Which bounded trial with synthetic records best tests the proposed approach?
- Run a bounded pilot with deliberately uncertain mock inputs
- Trial one region before enabling the next fictional region
- Run a synthetic utilization forecast with unexpected traffic growth
- Test mock publicity materials with one misleading caption
Explanation
Correct Answer
A:
Users have no transparent examples of error correction or escalation. Run a bounded pilot with deliberately uncertain mock inputs.
Incorrect Answers
B:
“Trial one region before enabling the next fictional region” tests a proposed change for the condition: Support capacity and user task quality have not been proven. This question instead requires an action that can rebuild trust through evidence, safeguards and iterative feedback. Users have no transparent examples of error correction or escalation. Run a bounded pilot with deliberately uncertain mock inputs.
C:
“Run a synthetic utilization forecast with unexpected traffic growth” tests a proposed change for the condition: Consumption can rise faster than the annual forecast. This question instead requires an action that can rebuild trust through evidence, safeguards and iterative feedback. Users have no transparent examples of error correction or escalation. Run a bounded pilot with deliberately uncertain mock inputs.
D:
“Test mock publicity materials with one misleading caption” tests a proposed change for the condition: The image may falsely imply a real situation occurred. This question instead requires an action that can rebuild trust through evidence, safeguards and iterative feedback. Users have no transparent examples of error correction or escalation. Run a bounded pilot with deliberately uncertain mock inputs.
Question 436
At Bridgewater Logistics, a legal search returns relevant documents below older irrelevant results. Evidence: Initial keyword matching is broad and top result quality is poor. The business goal is to consider semantic ranking after initial retrieval. The project owner requires a measurable acceptance criterion before signoff. Which evidence would most convincingly demonstrate that the requirement has been met?
- Answers identify the right clause or explicitly express uncertainty
- Outputs match the requested structure without copying example facts
- Highly relevant authorized sections rise without unacceptable delays
- Only permitted users see audio and transcript-derived outputs
Explanation
Correct Answer
C:
Initial keyword matching is broad and top result quality is poor. Highly relevant authorized sections rise without unacceptable delays.
Incorrect Answers
A:
“Answers identify the right clause or explicitly express uncertainty” validates an outcome in a case where the evidence is: Retrieval produced a plausible paragraph but not supporting evidence. This question instead requires an action that can consider semantic ranking after initial retrieval. Initial keyword matching is broad and top result quality is poor. Highly relevant authorized sections rise without unacceptable delays.
B:
“Outputs match the requested structure without copying example facts” validates an outcome in a case where the evidence is: Written instructions alone yield inconsistent section order. This question instead requires an action that can consider semantic ranking after initial retrieval. Initial keyword matching is broad and top result quality is poor. Highly relevant authorized sections rise without unacceptable delays.
D:
“Only permitted users see audio and transcript-derived outputs” validates an outcome in a case where the evidence is: Recording contains names and confidential discussions. This question instead requires an action that can consider semantic ranking after initial retrieval. Initial keyword matching is broad and top result quality is poor. Highly relevant authorized sections rise without unacceptable delays.
Question 437
At Prairie Health, a deployment manager assumes every Microsoft 365 subscriber has premium Copilot features. Evidence: Feature entitlements vary by license, tenant and rollout. The business goal is to validate actual licensing before promising a capability. Only non-sensitive synthetic data and approved sandbox systems may be used. Which bounded trial with synthetic records best tests the proposed approach?
- Edit a fictional plan with two sandbox colleagues
- Model fictional seat cohorts with different utilization and quality outcomes
- Test mock restricted files under permitted and denied identities
- Compare access using two sanctioned test-license profiles
Explanation
Correct Answer
D:
Feature entitlements vary by license, tenant and rollout. Compare access using two sanctioned test-license profiles.
Incorrect Answers
A:
“Edit a fictional plan with two sandbox colleagues” tests a proposed change for the condition: Parallel copies cause conflicting versions of one plan. This question instead requires an action that can validate actual licensing before promising a capability. Feature entitlements vary by license, tenant and rollout. Compare access using two sanctioned test-license profiles.
B:
“Model fictional seat cohorts with different utilization and quality outcomes” tests a proposed change for the condition: License usage is not equivalent to measurable task improvement. This question instead requires an action that can validate actual licensing before promising a capability. Feature entitlements vary by license, tenant and rollout. Compare access using two sanctioned test-license profiles.
C:
“Test mock restricted files under permitted and denied identities” tests a proposed change for the condition: Index query results lack caller-specific authorization filters. This question instead requires an action that can validate actual licensing before promising a capability. Feature entitlements vary by license, tenant and rollout. Compare access using two sanctioned test-license profiles.
Question 438
At Monarch Energy, a retail chatbot expects sharp holiday peaks. Evidence: Quota limits and token throughput may prevent desired traffic. The business goal is to plan capacity and graceful degradation around measured quotas. 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?
- Find lower-trust instructions in retrieved passages
- Calculate peak token rates, service limits and fallback needs
- Enumerate all workload components and usage assumptions
- Determine which connected entities and files the analyst can access
Explanation
Correct Answer
B:
Quota limits and token throughput may prevent desired traffic. Calculate peak token rates, service limits and fallback needs.
Incorrect Answers
A:
“Find lower-trust instructions in retrieved passages” is a diagnostic step for the reported condition: The document includes text telling the assistant to export hidden records. This question instead requires an action that can plan capacity and graceful degradation around measured quotas. Quota limits and token throughput may prevent desired traffic. Calculate peak token rates, service limits and fallback needs.
C:
“Enumerate all workload components and usage assumptions” is a diagnostic step for the reported condition: Indexing, embeddings, storage, evaluation and idle hosting also cost money. This question instead requires an action that can plan capacity and graceful degradation around measured quotas. Quota limits and token throughput may prevent desired traffic. Calculate peak token rates, service limits and fallback needs.
D:
“Determine which connected entities and files the analyst can access” is a diagnostic step for the reported condition: People, files and collaboration events have permissioned relationships. This question instead requires an action that can plan capacity and graceful degradation around measured quotas. Quota limits and token throughput may prevent desired traffic. Calculate peak token rates, service limits and fallback needs.
Question 439
At BrightPath Media, an audit department queries a policy assistant that cites a superseded manual. Evidence: The knowledge index was last refreshed before the recent policy revision. The business goal is to maintain fresh retrieval sources rather than changing model weights. The proposed solution must respect documented authority and organizational controls. Which approach most directly meets the stated business objective within the approved authority?
- Use original brand-approved prompts and human creative review
- Draft from approved clauses and route final language to legal review
- Add quality checks and structured table parsing for critical documents
- Set ingestion and retirement rules for approved policy documents
Explanation
Correct Answer
D:
The knowledge index was last refreshed before the recent policy revision. Set ingestion and retirement rules for approved policy documents. A policy that changes after indexing will not be reflected in retrieved passages until the source is refreshed. Track source versions, ingestion success and index freshness instead of retraining the model for factual updates.
Incorrect Answers
A:
“Use original brand-approved prompts and human creative review” is a design or implementation response to the condition: Similar branded content may create legal or reputational risks. This question instead requires an action that can maintain fresh retrieval sources rather than changing model weights. The knowledge index was last refreshed before the recent policy revision. Set ingestion and retirement rules for approved policy documents. A policy that changes after indexing will not be reflected in retrieved passages until the source is refreshed. Track source versions, ingestion success and index freshness instead of retraining the model for factual updates.
B:
“Draft from approved clauses and route final language to legal review” is a design or implementation response to the condition: The proposed wording could create commitments that the legal owner has not approved. This question instead requires an action that can maintain fresh retrieval sources rather than changing model weights. The knowledge index was last refreshed before the recent policy revision. Set ingestion and retirement rules for approved policy documents. A policy that changes after indexing will not be reflected in retrieved passages until the source is refreshed. Track source versions, ingestion success and index freshness instead of retraining the model for factual updates.
C:
“Add quality checks and structured table parsing for critical documents” is a design or implementation response to the condition: Corrupt source extraction can make even a grounded answer wrong. This question instead requires an action that can maintain fresh retrieval sources rather than changing model weights. The knowledge index was last refreshed before the recent policy revision. Set ingestion and retirement rules for approved policy documents. A policy that changes after indexing will not be reflected in retrieved passages until the source is refreshed. Track source versions, ingestion success and index freshness instead of retraining the model for factual updates.
Question 440
At Coral Bay Systems, an IT leader wants to renew hundreds of Copilot seats based only on logins. Evidence: License usage is not equivalent to measurable task improvement. The business goal is to evaluate per-seat productivity and quality before procurement. The proposed solution must respect documented authority and organizational controls. Which approach most directly meets the stated business objective within the approved authority?
- Pilot renewal tiers based on observed task-level benefit and governance
- Use hybrid retrieval with suitable ranking and filters
- Separate web and work findings with dated citations
- Use supported Teams Copilot to extract decisions and open questions
Explanation
Correct Answer
A:
License usage is not equivalent to measurable task improvement. Pilot renewal tiers based on observed task-level benefit and governance.
Incorrect Answers
B:
“Use hybrid retrieval with suitable ranking and filters” is a design or implementation response to the condition: Vector-only matching risks missing literal identifiers. This question instead requires an action that can evaluate per-seat productivity and quality before procurement. License usage is not equivalent to measurable task improvement. Pilot renewal tiers based on observed task-level benefit and governance.
C:
“Separate web and work findings with dated citations” is a design or implementation response to the condition: Public web claims and workplace records have different provenance. This question instead requires an action that can evaluate per-seat productivity and quality before procurement. License usage is not equivalent to measurable task improvement. Pilot renewal tiers based on observed task-level benefit and governance.
D:
“Use supported Teams Copilot to extract decisions and open questions” is a design or implementation response to the condition: Messages contain proposals, approvals and timestamps. This question instead requires an action that can evaluate per-seat productivity and quality before procurement. License usage is not equivalent to measurable task improvement. Pilot renewal tiers based on observed task-level benefit and governance.