1. Which four signals mark a promising AI candidate?○ Executive sponsorship, budget, vendor availability, and a deadline○ Volume, repetition of judgement, a recorded outcome, and tolerable error○ High cost, high visibility, strategic importance, and competitive pressure○ Structured data, cloud hosting, a large dataset, and a data scientist
2. Which of the four signals is most often absent and least often checked?○ Volume○ Repetition of judgement○ A recorded outcome○ Tolerable error
3. Why is the tedium round phrased as "what did you do more than ten times last week that a bright new joiner could have done after two days' training"?○ To identify training gaps in the team○ Because asking where AI could help returns what people have read about, and asking what is difficult returns the hardest tasks — which are the worst candidates; the aim is to surface dull work people do not volunteer○ To measure individual productivity○ To determine which roles could be made redundant
4. What is the purpose of the wish round?○ To gather feature requests for existing systems○ To surface new-capability candidates, which do not appear in the other rounds because people do not think of absent work as work○ To assess team morale○ To prioritise the candidates already found
5. Which are legitimate paper-trail sources of use cases? Select all that apply.☐ Complaint and escalation logs☐ Rework and correction records☐ Handover documents written before someone goes on leave☐ Competitor press releases
6. An executive returns from a conference insisting on a specific AI project. What is the productive response?○ Refuse on the grounds that the idea did not come from a structured process○ Implement it as specified to preserve the relationship○ Translate it — propose the version that would work here because the data exists, keeping the sponsorship while replacing the project○ Escalate the disagreement to the board
7. Why does customer-facing AI need a tighter exception path than internal AI?○ Because customer-facing models are inherently less accurate○ Because a mistake is visible outside the organisation, may be public, and may be regulated○ Because customer-facing systems process more data○ Because internal users are more tolerant of downtime
8. What makes agent-assist a better first customer-service project than full automation?○ It requires no integration work○ Accountability stays with a person, and each agent edit is a labelled example showing what the system proposed against what a competent human sent — the evidence needed to automate specific categories later○ It is always cheaper to license○ Customers always prefer speaking to humans
9. Which HR use case carries the highest risk?○ Answering staff questions from published HR policies○ Drafting job descriptions for human sign-off○ Scoring, ranking or filtering job applicants○ Answering routine onboarding questions
10. What is the document-in, data-out pattern, and why is it the highest-yield general candidate?○ Converting paper records to digital storage; it is cheap to implement○ A document arrives, a person extracts fields and makes a routine judgement, and the result is entered into a system — high volume, repeating judgement, errors that reconcile against something else, disliked work so adoption is easy, and every historical document already carries its answer key○ Generating documents automatically from structured data; it eliminates writing time○ Storing documents in a searchable index; it improves retrieval speed