7 Prompt Habits That Fix Bad AI Answers

When an AI gives you a bland, generic answer, it's usually doing exactly what you asked. You asked a bland, generic question.

Think of it like briefing a very smart new colleague who started this morning. They know a lot about the world and nothing about you, your project or what "good" looks like to you. Everything below is just ways of giving them that missing context.

1. Say who it's for

"Explain inflation" gets you a textbook paragraph. "Explain inflation to my 12-year-old, who just asked why her favourite snack costs more" gets you something you can actually use.

The audience changes everything: vocabulary, length, examples, tone. It's the single fastest upgrade to any prompt.

2. Show what you've got

Don't describe your draft. Paste it. Don't summarize the error. Paste the whole error message. Don't say "a formal email". Paste an email you've written before that sounds right.

AI is far better at working with real material than at guessing what you meant.

3. Ask for the format you want

If you want a table, ask for a table. Want three options instead of one? Say so. Need it under 100 words for a LinkedIn post? Put the number in.

A few that work well:

• "Give me three versions: formal, friendly and very short."
• "Answer in a table with columns for price, pros and cons."
• "Start with a one-sentence answer, then explain."

4. Tell it what to avoid

This one is underrated. If every answer you get starts with "Great question!" or ends with a wall of disclaimers, tell it not to. If you hate corporate buzzwords, list the ones that make you wince.

"No intro, no summary at the end, and don't use the words 'leverage' or 'seamless'."

5. Let it ask you questions

For anything bigger than a quick question, end your prompt with:

"Before you start, ask me any questions you need answered to do this well."

You'll be surprised how often it asks about something you forgot to mention: the budget, the deadline, who has to approve it. Thirty seconds answering beats three rounds of corrections.

6. Break big jobs into steps

"Write my business plan" produces something long and hollow. "First, list the ten questions a bank would ask about my bakery business plan" produces something you can work with. Then do the next step. Then the next.

You stay in control, and you catch wrong turns early instead of on page nine.

7. Push back

The first answer is a draft, not a verdict. Good follow-ups:

• "That's too generic. Use a real example from a small business."
• "Shorter. Cut it by half."
• "What would someone who disagrees with this say?"
• "Which part of that are you least sure about?"

That last one is worth memorizing. AI can sound equally confident about things it knows and things it's guessing, and asking directly often surfaces the shaky bits.

A before-and-after

Before:

Write a cover letter for a marketing job.

After:

I'm applying for a junior marketing coordinator role at a small outdoor gear shop. I've run the Instagram for my university hiking club for two years (grew it from 300 to 2,100 followers). Write a cover letter under 250 words. Friendly but not cheesy. Mention the hiking club early. No "I am writing to express my interest".

Same tool, same model, wildly different result. The second one took maybe 40 seconds longer to write.

The one-line version

If you remember nothing else: give it the context a new colleague would need. Who it's for, what you've got, what good looks like and what to avoid.

Want to see how the main chatbots compare once you're prompting well? Read ChatGPT vs Claude vs Gemini.