STAGE 01 · PART A · LESSON 04 OF 20

Make a Good Answer Great

Refine weak answers using feedback, comparisons and deliberate constraints.

1. Treat the first answer like a rough draft

Even a good prompt won’t always get the result you want. That’s normal. Editing with AI is a skill: you notice what’s missing, name it precisely and ask for a second attempt.

Think of three quality questions: Is it relevant? Is it clear? Can I actually use it? If the answer fails one, tell the AI exactly how.

2. Give feedback the way you would to a colleague

Instead of “That’s rubbish”, try “This is too formal for a friendly message. Keep the same meaning, remove the jargon and use shorter sentences.” Specific corrections usually work better than vague dissatisfaction.

This is a good start. Make three changes: (1) reduce it to 150 words, (2) replace technical terms with plain English, and (3) add one concrete example. Keep the original facts unchanged.

3. Ask for options, then choose

When you’re not sure which direction is best, ask for two or three noticeably different approaches. For instance: “Give me a warm version, a direct version and a short version.” Then choose the one that fits and explain why.

It can also help to ask the AI to compare options against criteria: cost, effort, risk, clarity or speed. Just remember that the scores reflect the information supplied; they are not independently verified facts.

4. Use the “critic then improve” technique

Ask: “What are the three biggest weaknesses in this answer? Revise it to address them.” You should still check the critique. The point is to encourage a second pass, not assume the second pass is perfect.

5. Know when to stop

Endless tweaking can waste time. Before you start, define success: “I want a polite message that’s ready for me to check and send.” Once you’ve achieved that, stop. A useful result beats a theoretically perfect result you never use.

YOUR TURN

Try it yourself

Ask the AI to draft a short fictional message inviting neighbours to a community litter pick. Then give four rounds of feedback: tone, length, clarity and format.

Write a friendly invitation to a neighbourhood clean-up. Make up no venue or date. Use placeholders for missing details.

Now ask for a version suitable for a printed flyer and one for a group chat. Note what changes and what must remain accurate.

KNOWLEDGE CHECK

Can you explain it back?

Answer each question before opening the explanation. You aren’t being graded — this is how you make the knowledge stick.

1. What’s more useful than “make it better”?

Specific feedback explaining what’s wrong and what to change.

2. Should you accept a revised answer without checking it?

No. Revisions can introduce new mistakes or alter meaning.

3. When should you stop refining?

When the output meets your clear, practical success criteria and you have checked it.

Remember: Good AI users are not just good at asking. They’re good at reviewing, correcting and stopping at useful.

Lesson 03 Lesson 05