Af AI Fluency Academy
Module 42%
Save & exit
Foundations · Module 2
Constrain
The cost of vague asks
Naming the format
Tell it what good looks like
Setting hard limits
Practice task
Module check
Lesson 3 of 6 · ~7 min

Tell the model what good looks like

Vague requests get vague results. When you ask an AI to "summarize this" or "clean this up," you are leaving every important decision to the model: how long, what to keep, what format, what to leave out. It will guess, and you will spend more time fixing the guess than you saved.

The fluent move is to state the exact shape of the output you need before you ask for it. You are not writing code and you are not learning secret words. You are giving clear direction, the same way you would brief a sharp new colleague who has no context.

Vague
"Summarize these customer interviews for me."
Clear
"Give me five bullet points, each one a recurring theme, with a short quote. No preamble."

The three things to name

Before you send a request, decide and state these three things. They take seconds and save minutes:

1.
Format. What shape should the answer take? A table, five bullets, a single paragraph, a JSON-free plain list.
2.
Limits. How long, how many, what scope? "At most 150 words." "Only the top three."
3.
Exclusions. What should never appear? "No marketing language." "Do not invent details that are not in the source."
Try it

Rewrite this vague request so it names a format, a limit, and one exclusion: "Write up the team offsite notes."

{{ draftHint }}
One strong version
"Turn the offsite notes into a one-page recap: a 3-sentence summary, then decisions as a bulleted list, then action items as a table with owner and due date. Keep it under 250 words and leave out anything not discussed."
← The cost of vague asks

This is optional preparation. The Applied AI Institute certification is assessed independently, and what you practice here is never the exam itself.