Microsoft AB-731 Practice Test 34
Question 330
At Ashwood Public Services, a program manager asks for outstanding decisions across permitted Teams and Word files. Evidence: The answer depends on authorized Microsoft 365 work context. The business goal is to choose an appropriately licensed work-grounded Copilot experience. 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?
- Calculate peak token rates, service limits and fallback needs
- List required read and write operations separately
- Check user permissions and which relevant files are accessible
- Count required sources and assess breadth, recency and permissions
Explanation
Correct Answer
C:
The answer depends on authorized Microsoft 365 work context. Check user permissions and which relevant files are accessible. Answers that rely on company mail, meetings and documents must use an authorized work-grounded experience with the appropriate tenant entitlements and permissions. Public-only answers cannot silently substitute private context.
Incorrect Answers
A:
“Calculate peak token rates, service limits and fallback needs” is a diagnostic step for the reported condition: Quota limits and token throughput may prevent desired traffic. This question instead requires an action that can choose an appropriately licensed work-grounded Copilot experience. The answer depends on authorized Microsoft 365 work context. Check user permissions and which relevant files are accessible. Answers that rely on company mail, meetings and documents must use an authorized work-grounded experience with the appropriate tenant entitlements and permissions. Public-only answers cannot silently substitute private context.
B:
“List required read and write operations separately” is a diagnostic step for the reported condition: An unnecessary permission increases risk without improving the task. This question instead requires an action that can choose an appropriately licensed work-grounded Copilot experience. The answer depends on authorized Microsoft 365 work context. Check user permissions and which relevant files are accessible. Answers that rely on company mail, meetings and documents must use an authorized work-grounded experience with the appropriate tenant entitlements and permissions. Public-only answers cannot silently substitute private context.
D:
“Count required sources and assess breadth, recency and permissions” is a diagnostic step for the reported condition: The question requires multi-step evidence gathering rather than quick lookup. This question instead requires an action that can choose an appropriately licensed work-grounded Copilot experience. The answer depends on authorized Microsoft 365 work context. Check user permissions and which relevant files are accessible. Answers that rely on company mail, meetings and documents must use an authorized work-grounded experience with the appropriate tenant entitlements and permissions. Public-only answers cannot silently substitute private context.
Question 331
At Silverpine Group, a company wants enterprise copilots before classifying sensitive shared files. Evidence: Unreviewed permissions could expose confidential material to broader employees. The business goal is to treat data governance as a prerequisite to AI adoption. 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?
- Map stakeholders, skills and handoffs for rollout stages
- Identify sponsor, business goals and decision rights
- Inventory missing model card and operational playbook sections
- Inspect labels, data owners, access groups and retention settings
Explanation
Correct Answer
D:
Unreviewed permissions could expose confidential material to broader employees. Inspect labels, data owners, access groups and retention settings.
Incorrect Answers
A:
“Map stakeholders, skills and handoffs for rollout stages” is a diagnostic step for the reported condition: IT, HR, security and functional leaders run disconnected activities. This question instead requires an action that can treat data governance as a prerequisite to AI adoption. Unreviewed permissions could expose confidential material to broader employees. Inspect labels, data owners, access groups and retention settings.
B:
“Identify sponsor, business goals and decision rights” is a diagnostic step for the reported condition: Employees receive mixed signals about priority and approved use. This question instead requires an action that can treat data governance as a prerequisite to AI adoption. Unreviewed permissions could expose confidential material to broader employees. Inspect labels, data owners, access groups and retention settings.
C:
“Inventory missing model card and operational playbook sections” is a diagnostic step for the reported condition: Users and auditors cannot see intended scope and failure conditions. This question instead requires an action that can treat data governance as a prerequisite to AI adoption. Unreviewed permissions could expose confidential material to broader employees. Inspect labels, data owners, access groups and retention settings.
Question 332
At Solstice Retail, a manager requests actionable tasks from a long team chat. Evidence: Messages contain proposals, approvals and timestamps. The business goal is to summarize only confirmed actions from authorized discussion. The project owner requires a measurable acceptance criterion before signoff. Which evidence would most convincingly demonstrate that the requirement has been met?
- The model separates supported visual facts from inferred causes
- The assistant informs but cannot unilaterally approve exceptions
- Unauthorized users cannot change stock and audit evidence is retained
- Actions retain correct owner and evidence instead of invented assignments
Explanation
Correct Answer
D:
Messages contain proposals, approvals and timestamps. Actions retain correct owner and evidence instead of invented assignments.
Incorrect Answers
A:
“The model separates supported visual facts from inferred causes” validates an outcome in a case where the evidence is: Text-only signals cannot describe the damage pattern. This question instead requires an action that can summarize only confirmed actions from authorized discussion. Messages contain proposals, approvals and timestamps. Actions retain correct owner and evidence instead of invented assignments.
B:
“The assistant informs but cannot unilaterally approve exceptions” validates an outcome in a case where the evidence is: Regulated interpretation requires designated decision-makers. This question instead requires an action that can summarize only confirmed actions from authorized discussion. Messages contain proposals, approvals and timestamps. Actions retain correct owner and evidence instead of invented assignments.
C:
“Unauthorized users cannot change stock and audit evidence is retained” validates an outcome in a case where the evidence is: Reading policy is different from invoking a high-impact connector action. This question instead requires an action that can summarize only confirmed actions from authorized discussion. Messages contain proposals, approvals and timestamps. Actions retain correct owner and evidence instead of invented assignments.
Question 333
At Harborline Products, a benefits team wants AI to approve employee exceptions automatically. Evidence: Decisions have legal consequences and policy ambiguity. The business goal is to limit automation to low-risk repeatable tasks and preserve human approvals. The proposed solution must respect documented authority and organizational controls. Which approach most directly meets the stated business objective within the approved authority?
- Automate standard policy retrieval but route exceptions to authorized reviewers
- Tune retrieval cutoff and ranking with agreed quality criteria
- Keep arithmetic controls in validated code and use AI for exception explanation
- Pilot a suitable pretrained model with safe prompting and evaluation
Explanation
Correct Answer
A:
Decisions have legal consequences and policy ambiguity. Automate standard policy retrieval but route exceptions to authorized reviewers. Routine policy retrieval can be automated, but discretionary legal or benefits exceptions require an authorized human decision.
Incorrect Answers
B:
“Tune retrieval cutoff and ranking with agreed quality criteria” is a design or implementation response to the condition: Retrieving many passages improves coverage while increasing noise. This question instead requires an action that can limit automation to low-risk repeatable tasks and preserve human approvals. Decisions have legal consequences and policy ambiguity. Automate standard policy retrieval but route exceptions to authorized reviewers. Routine policy retrieval can be automated, but discretionary legal or benefits exceptions require an authorized human decision.
C:
“Keep arithmetic controls in validated code and use AI for exception explanation” is a design or implementation response to the condition: The reconciliation is deterministic arithmetic with known control rules. This question instead requires an action that can limit automation to low-risk repeatable tasks and preserve human approvals. Decisions have legal consequences and policy ambiguity. Automate standard policy retrieval but route exceptions to authorized reviewers. Routine policy retrieval can be automated, but discretionary legal or benefits exceptions require an authorized human decision.
D:
“Pilot a suitable pretrained model with safe prompting and evaluation” is a design or implementation response to the condition: No task-specific labeled training data has yet been collected. This question instead requires an action that can limit automation to low-risk repeatable tasks and preserve human approvals. Decisions have legal consequences and policy ambiguity. Automate standard policy retrieval but route exceptions to authorized reviewers. Routine policy retrieval can be automated, but discretionary legal or benefits exceptions require an authorized human decision.
Question 334
At Crownridge Travel, an internal chatbot occasionally produces offensive responses to provocative inputs. Evidence: Unmoderated content could harm employees or customers. The business goal is to apply proportionate output safety checks and escalation. The proposed solution must respect documented authority and organizational controls. Which approach most directly meets the stated business objective within the approved authority?
- Rotate compromised keys and use managed identity or vault patterns
- Build a multi-year TCO comparison with sensitivity ranges
- Add safety evaluation, filtering and human escalation for high risk
- Use original brand-approved prompts and human creative review
Explanation
Correct Answer
C:
Unmoderated content could harm employees or customers. Add safety evaluation, filtering and human escalation for high risk. Potentially harmful outputs need safety screening suited to the audience and use case. Test borderline cases, provide escalation routes and log policy failures without treating a classifier as infallible.
Incorrect Answers
A:
“Rotate compromised keys and use managed identity or vault patterns” is a design or implementation response to the condition: The credentials enable unauthorized paid inference requests. This question instead requires an action that can apply proportionate output safety checks and escalation. Unmoderated content could harm employees or customers. Add safety evaluation, filtering and human escalation for high risk. Potentially harmful outputs need safety screening suited to the audience and use case. Test borderline cases, provide escalation routes and log policy failures without treating a classifier as infallible.
B:
“Build a multi-year TCO comparison with sensitivity ranges” is a design or implementation response to the condition: Integration, human review, logging and employee training are omitted. This question instead requires an action that can apply proportionate output safety checks and escalation. Unmoderated content could harm employees or customers. Add safety evaluation, filtering and human escalation for high risk. Potentially harmful outputs need safety screening suited to the audience and use case. Test borderline cases, provide escalation routes and log policy failures without treating a classifier as infallible.
D:
“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 apply proportionate output safety checks and escalation. Unmoderated content could harm employees or customers. Add safety evaluation, filtering and human escalation for high risk. Potentially harmful outputs need safety screening suited to the audience and use case. Test borderline cases, provide escalation routes and log policy failures without treating a classifier as infallible.
Question 335
At Ashwood Public Services, a regional chain proposes an enterprise-wide copilot rollout before testing. Evidence: High deployment effort conceals whether one use case can succeed. The business goal is to use a bounded high-value pilot with explicit stop criteria. The operating team needs an ongoing check after the initial rollout. Which post-rollout signal is most important for detecting regression or governance drift?
- Watch error drift and disproportionate escalation cost
- Monitor scope creep and benefit erosion during scale-up
- Watch coverage gaps and false confidence from synthetic-only metrics
- Monitor post-update regressions and reemerging old errors
Explanation
Correct Answer
B:
High deployment effort conceals whether one use case can succeed. Monitor scope creep and benefit erosion during scale-up. A bounded, measured pilot tests value, feasibility and risk before resources are committed to enterprise-wide expansion.
Incorrect Answers
A:
“Watch error drift and disproportionate escalation cost” monitors this signal: Simple tags may be handled by a lower-cost capable model. This question instead requires an action that can use a bounded high-value pilot with explicit stop criteria. High deployment effort conceals whether one use case can succeed. Monitor scope creep and benefit erosion during scale-up. A bounded, measured pilot tests value, feasibility and risk before resources are committed to enterprise-wide expansion.
C:
“Watch coverage gaps and false confidence from synthetic-only metrics” monitors this signal: Synthetic samples may reproduce omissions and hide rare real-world issues. This question instead requires an action that can use a bounded high-value pilot with explicit stop criteria. High deployment effort conceals whether one use case can succeed. Monitor scope creep and benefit erosion during scale-up. A bounded, measured pilot tests value, feasibility and risk before resources are committed to enterprise-wide expansion.
D:
“Monitor post-update regressions and reemerging old errors” monitors this signal: Transaction behavior changed since training. This question instead requires an action that can use a bounded high-value pilot with explicit stop criteria. High deployment effort conceals whether one use case can succeed. Monitor scope creep and benefit erosion during scale-up. A bounded, measured pilot tests value, feasibility and risk before resources are committed to enterprise-wide expansion.
Question 336
At Summit Manufacturing, a released agent sends misleading guidance to many employees. Evidence: There is no owner or process for AI-specific incidents. The business goal is to establish AI incident detection, containment and reporting. 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 tabletop exercise on a fictional incorrect-policy incident
- Test a fictional disputed eligibility decision with review steps
- Create a fictional system card for an FAQ assistant
- Run a synthetic utilization forecast with unexpected traffic growth
Explanation
Correct Answer
A:
There is no owner or process for AI-specific incidents. Run a tabletop exercise on a fictional incorrect-policy incident.
Incorrect Answers
B:
“Test a fictional disputed eligibility decision with review steps” tests a proposed change for the condition: No human owner accepted responsibility for the decision. This question instead requires an action that can establish AI incident detection, containment and reporting. There is no owner or process for AI-specific incidents. Run a tabletop exercise on a fictional incorrect-policy incident.
C:
“Create a fictional system card for an FAQ assistant” tests a proposed change for the condition: Users and auditors cannot see intended scope and failure conditions. This question instead requires an action that can establish AI incident detection, containment and reporting. There is no owner or process for AI-specific incidents. Run a tabletop exercise on a fictional incorrect-policy incident.
D:
“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 establish AI incident detection, containment and reporting. There is no owner or process for AI-specific incidents. Run a tabletop exercise on a fictional incorrect-policy incident.
Question 337
At Monarch Energy, a maintenance operator has images and brief text notes of equipment defects. Evidence: Text descriptions alone omit visual indicators needed for triage. The business goal is to evaluate vision-capable AI when image evidence materially matters. Only non-sensitive synthetic data and approved sandbox systems may be used. Which bounded trial with synthetic records best tests the proposed approach?
- Trial routine and exception cases using fictional benefits rules
- Test a mock legal summary alongside a restricted full contract
- Label fictional tickets including ambiguous mixed-urgency cases
- Test labeled fictional defect photos and paired notes
Explanation
Correct Answer
D:
Text descriptions alone omit visual indicators needed for triage. Test labeled fictional defect photos and paired notes. A textual description cannot confirm damage, layout or other cues visible only in an image. Evaluate an approved vision-capable system on labeled examples and keep ambiguous findings for human review.
Incorrect Answers
A:
“Trial routine and exception cases using fictional benefits rules” tests a proposed change for the condition: Decisions have legal consequences and policy ambiguity. This question instead requires an action that can evaluate vision-capable AI when image evidence materially matters. Text descriptions alone omit visual indicators needed for triage. Test labeled fictional defect photos and paired notes. A textual description cannot confirm damage, layout or other cues visible only in an image. Evaluate an approved vision-capable system on labeled examples and keep ambiguous findings for human review.
B:
“Test a mock legal summary alongside a restricted full contract” tests a proposed change for the condition: The user can access the approved summary but not full agreement text. This question instead requires an action that can evaluate vision-capable AI when image evidence materially matters. Text descriptions alone omit visual indicators needed for triage. Test labeled fictional defect photos and paired notes. A textual description cannot confirm damage, layout or other cues visible only in an image. Evaluate an approved vision-capable system on labeled examples and keep ambiguous findings for human review.
C:
“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 evaluate vision-capable AI when image evidence materially matters. Text descriptions alone omit visual indicators needed for triage. Test labeled fictional defect photos and paired notes. A textual description cannot confirm damage, layout or other cues visible only in an image. Evaluate an approved vision-capable system on labeled examples and keep ambiguous findings for human review.
Question 338
At Blue Ridge Technology, a transformation team receives dozens of unrelated AI requests. Evidence: No process ranks ideas by value, readiness and risk. The business goal is to govern intake and prioritization for sustainable AI investment. Only non-sensitive synthetic data and approved sandbox systems may be used. Which bounded trial with synthetic records best tests the proposed approach?
- Score three fictional proposals with conflicting benefits and hazards
- Submit a fictional wrong answer and trace it to resolution
- Run a bounded pilot with deliberately uncertain mock inputs
- Simulate a model change with one missing reviewer approval
Explanation
Correct Answer
A:
No process ranks ideas by value, readiness and risk. Score three fictional proposals with conflicting benefits and hazards.
Incorrect Answers
B:
“Submit a fictional wrong answer and trace it to resolution” tests a proposed change for the condition: Issues are visible only to individual users and are not triaged. This question instead requires an action that can govern intake and prioritization for sustainable AI investment. No process ranks ideas by value, readiness and risk. Score three fictional proposals with conflicting benefits and hazards.
C:
“Run a bounded pilot with deliberately uncertain mock inputs” tests a proposed change for the condition: Users have no transparent examples of error correction or escalation. This question instead requires an action that can govern intake and prioritization for sustainable AI investment. No process ranks ideas by value, readiness and risk. Score three fictional proposals with conflicting benefits and hazards.
D:
“Simulate a model change with one missing reviewer approval” tests a proposed change for the condition: No second approver checks the release evidence. This question instead requires an action that can govern intake and prioritization for sustainable AI investment. No process ranks ideas by value, readiness and risk. Score three fictional proposals with conflicting benefits and hazards.
Question 339
At Willow Creek Finance, 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 sponsor needs an evidence-based diagnosis rather than a guessed fix. Which initial investigation best identifies the controlling constraint before any solution is chosen?
- Map permitted actions, limits and approver identities
- Diagnose entitlement, sign-in, app version and tenant policy
- Inspect who can change model, evaluation gates and production permissions
- Identify trust zones, secrets, injection vectors and privileged actions
Explanation
Correct Answer
A:
The agent's action scope could exceed its intended advisory role. Map permitted actions, limits and approver identities. 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
B:
“Diagnose entitlement, sign-in, app version and tenant policy” is a diagnostic step for the reported condition: Users do not know whether the issue is access, training or configuration. 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. Map permitted actions, limits and approver identities. 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:
“Inspect who can change model, evaluation gates and production permissions” is a diagnostic step for the reported condition: No second approver checks the release evidence. 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. Map permitted actions, limits and approver identities. 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.
D:
“Identify trust zones, secrets, injection vectors and privileged actions” is a diagnostic step for the reported condition: The agent can call tools and access internal data repositories. 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. Map permitted actions, limits and approver identities. 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 340
At Solstice Retail, 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. The proposed solution must respect documented authority and organizational controls. Which approach most directly meets the stated business objective within the approved authority?
- Add quality checks and structured table parsing for critical documents
- Index embeddings for matching and use a generator only after retrieval
- Select a generative text model for reviewed descriptions and reserve prediction for forecasts
- Route basic classification to an evaluated efficient model and complex exceptions upward
Explanation
Correct Answer
B:
Similarity matching is distinct from composing a final recommendation. Index embeddings for matching and use a generator only after retrieval.
Incorrect Answers
A:
“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 distinguish embeddings for retrieval from LLMs for generation. Similarity matching is distinct from composing a final recommendation. Index embeddings for matching and use a generator only after retrieval.
C:
“Select a generative text model for reviewed descriptions and reserve prediction for forecasts” is a design or implementation response to the condition: The deliverable is new natural-language copy, not a numeric sales prediction. 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. Index embeddings for matching and use a generator only after retrieval.
D:
“Route basic classification to an evaluated efficient model and complex exceptions upward” is a design or implementation response to the condition: Simple tags may be handled by a lower-cost capable model. 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. Index embeddings for matching and use a generator only after retrieval.