Final test & certificate

Choose the best answer for each question.

To pass: at least 8 of 10 correct.

1. Question 1. What is the most accurate way to understand a 'prompt'?

   AA fixed magic spell for every caseBA work brief you give AI — the clearer, the better the resultCA line of programming code only IT can write

2. Question 2. What is the 4-part frame of a good prompt?

   ARole – Task – Context – FormatBMorning – Noon – Afternoon – EveningCInput – Process – Output – Save

3. Question 3. Why does HR especially benefit from prompting skills?

   ABecause HR needs to know codingBBecause most HR work is language (JDs, policy, email, interviews) — right in AI's wheelhouseCBecause AI fully replaces HR

4. Question 4. What does 'few-shot' mean in prompting?

   AWrite a very short promptBGive AI 1–2 sample examples of the desired output for it to mimicCOnly ask AI a few times then stop

5. Question 5. For a complex analytical task, which technique makes results more accurate and verifiable?

   AAsk AI to answer very fastBAsk AI to 'think step by step before concluding'CGive no context at all

6. Question 6. What is the right way to understand prompting?

   AMust be perfect on the first shotBA multi-round conversation: get a draft, then refineCUse one prompt for everything

7. Question 7. Which is a useful 'guardrail' to set in HR prompts?

   A'Feel free to invent figures to make it lively'B'Use only the information I provide, don't invent; if missing, write [to be added]'C'Write as long as possible'

8. Question 8. Which should NOT be pasted into a public AI tool?

   AA public sample JDBA payroll sheet with full names, ID numbers, and each person's salaryCA public company policy passage

9. Question 9. When you need AI to analyze real salary data, what is the right approach?

   APaste the raw identified dataBAnonymize first (Employee A, B…), drop IDs, keep only what's needed for the analysisCNever analyze salary at all

10. Question 10. What is the human's role in using AI for HR?

    AHand all HR decisions to AIBReview, verify, and be accountable for the final decisionCNo need to re-read AI output