Why Vague Prompts Produce Vague Output
Unit ID: AIT-M02-U00 Estimated active time: 15-20 minutes
The problem with vague requests
A vague prompt gives the AI tool very little to work with.
For example:
Write an email.
The tool does not know:
- who the email is for;
- why you are writing;
- what tone you need;
- what facts are safe to include;
- how long the email should be;
- what the reader should do next;
- what should be checked before sending.
So the tool guesses.
Sometimes the guess is useful. Sometimes it is wrong, too formal, too friendly, too long, too strong, or missing the real point.
A better way to think
Before asking the tool, pause and answer:
What would I tell a helpful human assistant?
You would not only say "write an email". You would explain the situation.
That explanation is the beginning of a prompt brief.
Weak prompt example
Weak prompt:
Make this better.
This is unclear. Better how?
- Shorter?
- More polite?
- Easier to understand?
- More confident?
- Less formal?
- More suitable for a manager?
- More suitable for a beginner?
The tool may improve the text in a way you did not want.
Stronger prompt example
Better prompt:
Rewrite this message for a colleague who is busy. Keep the tone polite and direct. Make it shorter than 120 words. Do not add new facts. Keep the request clear. After rewriting, list anything you assumed.
This is stronger because it gives:
- audience;
- tone;
- length;
- boundary;
- output requirement;
- review help.
Vague prompts can create risk
Vague prompts do not only waste time. They can create risk.
If the tool does not know the boundary, it may:
- invent details;
- reveal too much;
- sound too certain;
- make unfair assumptions;
- give advice outside your role;
- choose a decision you should make yourself.
Clear instructions reduce risk. They do not remove the need for review.
Quick practice
Read this weak prompt:
Summarise this.
List five missing details.
Then write a better version for a safe, nonsensitive article or note.