A vague prompt is a question to the training data. Compare vague, specific, and critically-revised prompts side by side to see what the model assumed when you didn't say.
Live preview · launch for the interactive version
See what a one-word prompt returns — the model's safest, most common guess.
Add detail and watch which assumptions get overridden, and which persist.
A critical rewrite surfaces what the default left out, and who it left out.
Predict the default for a vague prompt like 'a doctor' or 'a home.'
Hold the subject; add a single specifying word.
What changed, what stayed, what only a rewrite fixed?
Say what the model assumed and where that assumption came from.