AI Value Strategy and Use-case Portfolio Review questions AI Value Strategy and Use-case Portfolio — Integrated Summary Investment decisions start with a defined work outcome and measurement denominator, not a tool. Leaders separate usage from service improvement, establish a baseline and compare a realistic non-AI alternative. Eligibility gates then prevent large benefits from compensating for unresolved data rights or missing accountability. Weighted scores rank eligible opportunities, but sensitivity to weights and competition for resources require reasoned judgment. The economic case includes costs for the same period, separates released time from cash savings and tests alternative scenarios. Staged funding buys information before a broad commitment. Sunk cost is not a reason to continue, and expected value is not a profit promise. A decision record states what was selected or excluded and which new evidence would change the choice.
Review questions Test your understanding with these questions. Marking and saved results are available with a free account.
Question 1 A service team times each message, but the trial aims to reduce end-to-end request resolution time. One request can contain several messages. Which change aligns measurement with the goal? Average active work time per message Measure time from request opening to correct resolution Separate customer waiting and work, reporting only their means Time the first-to-last response without checking resolution status Question 2 Three opportunities pass gates and score well. All require the same data engineer in month one, and splitting that person's time would cause material delay. What comes next? Start all projects by splitting the engineer's time equally Approve based only on score order Fund the highest scores and resolve contention when it occurs Sequence work or secure capacity before portfolio approval Question 3 An unrecoverable amount was spent on a failed trial. From now on, continuing is expected to lose more than exiting, and both options meet safety conditions. What should govern the economic decision? Compare future flows from continuing and exiting Spread past expenditure over an extended trial period Continue until returns match expenditure already incurred Consider previous work duration Question 4 For a launch decision, security supplies an assessment before the decision, while customer service receives the approved decision afterward. Which RACI assignment fits these two roles? Security R for assessment; customer service C for approval Security A for launch; customer service I for team notification Security C for consultation; customer service I for notification Security I for awareness; customer service R for launch execution Question 5 An ordinal risk matrix assigns one risk a score of 4 and another 2, without financial or probability calibration. Which conclusion is justified? Rank the first higher, leaving harm magnitude unspecified Double the first risk's mitigation budget Estimate twice the expected loss when usage volumes match Convert the score sum into a common portfolio-loss estimate Question 6 Unsupported cases are referred to reviewers, but their arrival rate persistently exceeds team capacity. What response best fits a bounded trial? Extend review time without a backlog limit Narrow scope or add qualified capacity while tracking the queue Sample referrals instead of reviewing all without reassessing risk Spread referrals across staff whose competence is unverified Question 7 The organization differentiates through permissions over internal knowledge, while several suppliers offer general models. Its team can build integration but cannot maintain a proprietary model. Which design merits initial testing? Build a proprietary model using temporary vendor setup support Buy a closed product and adapt all knowledge to default permissions Wait for capability to maintain every layer Buy a replaceable model and build knowledge and permissions Question 8 A superseded policy was copied yesterday into a new file, giving the file a recent timestamp. Which evidence should determine retrieval eligibility? Latest modification date in the organization's file store File publication date without consulting approval status Version approval status with the responsible source Document ranking by visits during the latest period Question 9 A vendor contract provides compensation after an incident but does not specify notification timing or evidence supplied. Which operational gap needs attention? Define notification, evidence and ownership for containment Increase compensation payable against the vendor's annual report Add periodic contract review without changing notification timing Let the vendor choose notification time after closing its inquiry Question 10 The system produces a polished summary of an exception request. The decision requires checking an original document and professional interpretation. Where should delegation stop? Approve a structured summary containing every request field Check clarity then act on the recommendation without the source Use summaries as support; retain specialist verification and decision Request a second system draft and execute the more complete one Question 11 Trainees report confidence in checking sources after a presentation. What is stronger evidence of skill transfer? Repeat a detailed confidence survey one week after the session Test new conflicting-source cases against a defined standard Repeat presentation examples until everyone achieves full marks Compare daily usage with its pretraining level Question 12 An employee reports increased review burden and fears the objection will harm their appraisal. Which response supports learning during transformation? Record the report without a response or review commitment Refer it only to the vendor without examining internal work design Resolve only that employee's burden without checking the wider pattern Protect feedback, examine burden and communicate the decision Question 13 Intervention and comparison groups have equal means in one pretrial month. Does this establish parallel trends? No; examine multiple earlier periods and context changes Yes, if baseline levels and sample sizes match in that month Yes, if the comparison mean stays fixed in only the next month No, but a larger post-intervention sample can replace the assumption Question 14 A new subgroup has errors in two of five cases. Which interpretation is most defensible? Treat 40% as stable because every observed case was checked Defer all action until subgroup size matches the larger group Treat it as a concerning unstable signal needing representative evidence Use the organizational rate instead until the trial ends Question 15 A pathway was suspended after a control failure. Average performance later improves without evidence that the cause was repaired. What is an appropriate resumption condition? Resume after a quiet monitoring period without control retesting Verify repair and tests, obtain approval and set a review date Resume gradually if averages improve despite an unresolved cause Rely on a general vendor report not addressing the specific incident Create a free account to mark your answers and save your progress.