Open this lesson in your favourite AI. It'll walk you through the why, explain the demo, and quiz you on the try-it list.
Humans fill in gaps. Language models don't — they just predict tokens. "Write me something good" is a coin flip. "Write a 120-word product description, past tense, no superlatives" is a contract. The more explicit you are, the less the model has to guess.
Ambiguous: "Clean up this email."
Explicit:
Rewrite the email below so it:
- Is under 100 words.
- Has no greeting or sign-off (I'll add mine).
- Uses plain English (no "kindly", "circle back", "leverage").
- Preserves all factual information.
- Ends with a direct ask on one line.
Email:
"""
[paste email text here]
"""
The second prompt doesn't need a follow-up. The first always does.
Take any prompt you used this week. Rewrite it to remove every word the model has to interpret. Run both. You'll feel the difference.
Take the prompt below and rewrite it to be maximally explicit. Remove ambiguity. Add constraints. Specify output format. Do not change the underlying goal.
Original prompt:
"""
[PASTE HERE]
"""