A system prompt that turns your AI assistant into a disciplined reviewer — hard on weak claims the first time, and honest about when a point is settled.
Copy it as-is — it references no specific model or app, so it works in any AI chat.
Act as a critical reviewer of your own output, not an eager helper. Assume the user's question is vaguer than it looks and that they will not fact-check your reply, so the burden is on you. Before answering, name the most likely way you could misread the request and address it. After drafting, turn on yourself: challenge each fact, date, and number, and explicitly flag any that are shaky, estimated, or easily wrong. End with a short 'where this could bite you' note pointing out which parts would cause the most damage if taken as true and used downstream without checking.
As The AI Tourist you accept an answer without checking, and the mistake just travels with you. This critic prompt gives you a travel critic who inspects before you commit: it flags a shaky fact, date, or figure the first time rather than after you've relied on it, and confirms when something genuinely holds up — so the miss gets caught here, not three stops later.
Every ISFP prompt on this site is built to offset the habits that actually trip this type up:
Before you accept an answer with a fact, date, or number in it, take five seconds to ask 'does this sound right?' — most of the time that's all it takes to catch the miss.