Write your response in your own notes, then compare it with the sample answer. This practice is separate from the scored exam and does not affect your certificate.
Practice 1
Invented learner: Priya runs a small bakery and wants to compare two assistants for drafting supplier emails. Write her three reasoning questions (a real task, a question from her field with a small false assumption, and a problem whose answer she already knows), say how she should word them for the two assistants, and say how she should mark each answer.
Compare with a sample answer
Real task: ask each assistant to draft a polite email chasing a late flour delivery from the order details she supplies. False assumption: ask each to plan her opening hours around a delivery she knows arrives only on alternate Fridays, while the question says it arrives every Friday, to see whether it notices and corrects this. Known answer: a small sum she has already worked out, such as the cost of twelve bags at a price she supplies. She puts all three questions to both assistants in exactly the same wording, and marks each answer plainly usable, usable with correction or not usable, judged against what she already knows, rather than scoring out of ten.
The sample uses the three question types from the exercise, keeps the wording identical for both candidates and uses the three plain labels instead of a score.
Practice 2
Invented learner: Tom summarised the same long report with two assistants. One gave a polished first reply but needed four attempts and heavy editing, and the other needed one attempt. Say what Tom should record for each, which one took less effort, and why a polished first reply can mislead.
Compare with a sample answer
Tom records how he accessed each assistant, the number of attempts each needed (four and one) and how much editing the first needed. The second took less time and effort because effort is counted per useful result, not by how polished the first reply looks. The first reply looked impressive but cost him three extra attempts and heavy editing, so its polish hid the real work. Effort is one weight among six, so he would still check the other factors before deciding.
The sample counts attempts and editing per useful result, prefers the option that took less effort, and keeps effort as one weight among six.
Practice 3
Invented learner: Mei works with scanned PDFs that contain tables and with spreadsheets. Describe the two tests from the context and files exercise she should run on one candidate, what she should check before the second test, and what she should note afterwards.
Compare with a sample answer
First she picks the longest document she knows well, uploads it and asks about a specific detail from the middle, then checks the reply against what she knows. Second she chooses one file type she really uses, such as a PDF with tables, and checks that the candidate accepts it at all before testing it properly, then asks for one exact figure or row. Afterwards she notes anything it missed, invented or simply could not accept, and compares that with what the spec page claimed.
The sample follows the exercise: a middle-detail question on a document she knows, an acceptance check before the file-type test, and notes on missed, invented or unaccepted material.
Practice 4
Invented learner: Sam wants to try a new assistant for non-sensitive drafting. Write a short checklist of what Sam should read or check about data handling and about latency before relying on it.
Compare with a sample answer
Data handling: open the current privacy or data settings, write in his own words whether chats may be used to improve the service, note whether he can switch that off and whether he has already done so, and read the privacy policy for what is kept and for how long and who else can see it. Latency: test the wait at the times of day he actually works rather than once when the service is quiet, remembering that harder requests usually take longer by design. He pastes no passwords, other people's private details or confidential material while testing.
The sample covers the improvement-use setting and switching it off, the policy questions, testing latency at working times, and keeping the test clean.
Practice 5
Invented learner: Alicia tested two assistants for writing her weekly report. Fill in one model-log row for her, using the course's columns, and write the decision and the trigger that will make her test again.
Compare with a sample answer
Date: today's date. Task: weekly report. Candidate and access route: Candidate A, chat app. Reasoning check: usable with correction. Attempts needed: two attempts. Context and files tested: held a twelve page document and found a detail from the middle. Latency note: fine at 9am and at 4pm. Privacy checked: setting checked, improvement use switched off. Decision: keep for this task. Trigger: test again if answers start to feel worse, if the plan or its terms change, if a new release arrives or if her work changes, and otherwise review in a few months.
The sample fills each column the course lists, records only checks actually made, and names triggers from the course along with the every-few-months habit.
Practice 6
Invented learner: Noor has three tasks: summarising a confidential contract, brainstorming slogan ideas quickly, and checking a very long report. For each task name the two or three factors she should weigh first, and say what goes wrong if she skips this step.
Compare with a sample answer
Confidential contract: data handling first, with reasoning ability as a second factor if the summary must be right. Quick slogan brainstorm: latency and effort first. Very long report: context length and reasoning ability first. She decides these top two or three factors before comparing anything, because otherwise the most impressive-sounding feature will decide for her instead of the needs of the task.
The sample matches the course's examples (confidential document, quick brainstorm, long report) and states the reason for deciding the factors before comparing.