Are You Sure About That? The Prompt That Changed How I Use AI

“Are you sure about that?”

This should have been the first thing I said to AI after every initial answer. Why? It gives better responses when I doubt it from the start. I’ve tried it on several prompts — and the difference is real.

Why Skepticism Works

AI models are trained to be helpful. That sometimes means they confidently fill gaps with plausible-sounding information. When you challenge an answer, you force the model to reconsider — to flag what it actually knows versus what it’s guessing.

The Prompt I Use

I’ve found two versions useful, depending on how much control I want.

Short version

Always flag uncertain or speculative claims with [uncertain] or phrases like “I’m not sure about this.” State well-established facts directly. Don’t inflate responses with loosely supported points.

Long version

When answering questions, clearly distinguish between what you know with confidence and what is uncertain. For well-established facts, state them directly. For anything speculative, poorly documented, or based on indirect reasoning, flag it explicitly using phrases like “I’m not certain”, “this is speculative”, or a marker like [uncertain]. Do not pad responses with loosely related points just to seem thorough. If you don’t know something, say so plainly.

Both versions do the same job. The short one fits in a system prompt easily. The long one leaves no ambiguity.

What Changes

Once you add this to your setup, you start seeing [uncertain] markers appear. You notice the model hedging more. You trust what’s left unhedged a little more.

It’s not that the AI suddenly becomes smarter. It’s that it stops pretending to be more confident than it is.

What’s your experience? Try it yourself and see what changes.