A timed assessment is ending, but several items still need careful review. How would you prioritize?

Prioritize careful work under a timer using the actual assessment rules. Explain how you would balance remaining items, consequential errors and submission time.

Optional 20 minute mock assessment

Try these five original fictional cases before reading the worked answers. They reuse the exercises in questions 5, 10, 11, 12 and 21; only the first round of questions 10 and 11 is included. This is a practice exercise, with no employer scoring rule or hiring prediction.

Suggested timing: Two minutes to read the instructions, three minutes per case and three minutes to check and finish: 20 minutes in total. The timer is optional. Use only the supplied inputs, take a first pass without worked answers or outside tools, and record anything you leave unfinished.

Submission: Keep five numbered sections in a text document. Provide every output each case requests, including reasons and status distinctions. You may revisit cases before the timer ends. Mark any unfinished check explicitly; saving your document is the completion step for this practice. The member answer contains a worked key and a 20-point self-check; the points identify practice gaps rather than an assessment pass mark.

Case 1 Sentiment labels

Use these project-specific labels:

• positive: clearly favorable, with no expressed unfavorable view.
• negative: clearly unfavorable, with no expressed favorable view.
• mixed: expresses both favorable and unfavorable views.
• unclear: the expressed sentiment cannot be determined from this text alone. Do not infer tone from imagined context.

Label each complete comment and give one short reason:

1. "The setup was easy and I love the result."
2. "It crashes every time I save. I'm disappointed."
3. "The screen looks great, but the battery life is terrible."
4. "Well, that happened."

These labels are an exercise contract, not a universal sentiment taxonomy.

Case 2 Facts and format

Fact sheet: The support desk opens at 09:00 on weekdays.

Request: "Give the weekday opening time as one JSON object with exactly one key, opens_at, and a string value."

Response A: "The support desk opens at 09:00 on weekdays."

Response B: {"opens_at": "08:00"}

Rubric: Record factual correctness and format compliance separately as pass/fail. A response is acceptable only when both pass. This exercise does not request a pairwise winner or specify weights. State each response's two ratings and whether it is acceptable.

Case 3 Concise rationale

Approved policy: Requests received within 14 days of purchase are eligible for a return. Requests received after 14 days are not eligible under this policy. There are no additional exceptions in the supplied policy.

User: "I purchased this item 18 days ago. Am I eligible for a return?"

Response A: "The policy permits returns within 14 days, so an 18-day-old purchase is not eligible under the supplied policy."

Response B: "You are eligible because the return window is 30 days."

Task: Prefer the response supported by the policy. Write a rationale of no more than two sentences identifying the decisive factual error; do not restate both full responses.

Case 4 Evidence mapping

Evaluate this four-claim summary of the Tarin Pilot Grant using only the approved packet. The fictional project uses guideline version 3. External research is not allowed, and the excerpts do not establish facts they omit.

Approved source packet:

• S1: Organizations with 5–50 employees, inclusive, meet the grant’s staff-size eligibility condition. This excerpt establishes only staff-size eligibility.
• S2: The maximum grant is $2,000.
• S3: Equipment purchases are permitted expenses. This excerpt provides no information about travel expenses or application review times.

AI response, split into claims:

• C1: An organization with 12 employees meets the staff-size eligibility condition.
• C2: The maximum grant is $3,000.
• C3: Applications are reviewed within seven days.
• C4: The grant can pay for equipment and travel expenses.

Allowed evidence_status labels:

• supported: every factual part is established by the packet, with no contradiction.
• contradicted: at least one factual part is explicitly incompatible with the packet. This takes precedence over partial or absent support.
• absent: no factual part is supported or contradicted by the packet.
• partial: at least one factual part is supported and another is unaddressed, with no contradiction.

Grounding acceptance rule: Mark the response acceptable only if every claim is supported; otherwise mark it not_acceptable. This is a grounding judgment. An absent claim is not thereby proven false.

Produce: One record per claim containing its ID, evidence status, decisive source IDs and a short reason. For absent evidence, use an empty decisive-source list and name the sources checked. For a partial claim, identify the supported and unaddressed parts. Then give the overall rating and a submitted rationale of no more than two sentences.

Case 5 Permitted assistance

This is a work-sample exercise, not an employer test. Guideline version 3 permits annotation assistance under this exact tool contract:

• Built-in Prelabel Assist 1.3 is approved inside the project workspace to suggest labels for these items. An annotator must inspect each source text and record a decision; suggestions cannot be bulk-submitted as verified labels.
• QuickLabel AI, an external service offered through a personal account, is not approved. No project text, screenshots or records may be uploaded to it.
• Keep the original suggestion, the chosen action, the resulting label and its reason in the approved project record. No submission has occurred yet.

Allowed labels:

• request_refund: an explicit, unnegated request to return the payment or issue a refund.
• other_request: an explicit request for a different action, with no unnegated refund request.
• needs_review: the text does not establish which of the two request labels applies. This routes the item for review; it does not certify that review is complete.

Source items and built-in suggestions:

• P1: “I want a refund for this order.” Original prelabel: request_refund.
• P2: “I do not want a refund; I only need a copy of the receipt.” Original prelabel: request_refund.
• P3: “Can we sort this out?” Original prelabel: other_request. No earlier conversation is supplied.

Produce: A tool-use decision for both services. For each item, retain the original prelabel and choose accept, correct or review, then give the resulting allowed label and source-based reason. accept and correct can be ready for normal submission; review remains pending review. Finish with a short audit note naming the guideline and approved tool version. Do not claim the items were submitted or that a pending item was reviewed.