Is this what people think it is?

When you encounter a claim, image, or piece of content online, your instinct is often to ask whether it's true. Mike Caulfield argues that's the wrong question. What usually matters is whether the thing is what people believe it to be, and that is a different problem.

The prompt "Is this what people think it is?" directs an AI with web search to examine not just the factual content of a claim, but its context, origin, and how it's being understood and shared. A statistic might be accurate but misattributed. A video might be real but from a different event. A study might exist but say something far more limited than the headline suggests.

What makes the prompt effective is its framing. Rather than asking the AI to render a verdict, it asks the AI to surface the gap between what something is and what it's taken to mean. That's a more nuanced question, and it tends to produce more useful answers: concise, contextualised, and focused on the specific way a piece of information is likely to mislead.

It works best as a follow-up after an initial check, once you have a rough sense of the claim. Paste in the content (text, a screenshot, a quoted statistic) and let the AI examine it against what's actually being claimed and believed.