Ask AI What It Is Assuming

AI often gives you a clean answer even when parts of the request are unclear. That is useful, but risky. A polished answer can hide guesses about your goal, audience, data, budget, timeline, or definition of success. Ask: what are you assuming?

Do not treat the answer as a confession of perfect uncertainty. Treat it as a review checklist. If AI says it assumed your audience is non-technical, your data is current, or your priority is speed, you now know exactly what to confirm before using the work.

This is strongest after summaries, recommendations, migration plans, reporting conclusions, and project scopes. The follow-up can be very short: List the assumptions behind this answer. Mark which ones I should verify. Combine it with Be Critical. No Compliments. when you want the weak parts faster.

Use this exact pattern: What are you assuming here? Give me 5 bullets and mark anything I should verify before acting. It works because it moves AI from answer mode into review mode. You are not asking for more text. You are asking for the hidden conditions behind the text.