Spotting fakes

If there's one lesson uniting misconceptions, the Mandela effect, and AI hallucinations, it's this: feeling true and being true are different things. Here's a practical toolkit for telling them apart—without becoming an insufferable cynic.

Why we fall for 'truthy'

• Fluency bias: smooth, confident, familiar-sounding claims feel more true — regardless of accuracy. It's why a good-sounding myth (or AI answer) slips past our guard;
• It fits the pattern: we accept things that match what we already expect — the monocle 'belongs' on the rich cartoon man (Mandela effect);
• Repetition breeds belief: the more we hear something, the truer it feels (the 'illusory truth effect') — and the internet repeats everything;
• Confidence is contagious: an assured tone reads as expertise, whether from a person or a chatbot.

The verification habits

1. Pause on the specifics: exact numbers, quotes, dates and named sources are where fabrication concentrates — the most impressive-sounding detail is the one most worth checking;
2. Go to the primary source: trace a claim to its origin, not to someone repeating it. 'A study found...' — which study? Read it (or at least confirm it exists);
3. Check whether cited sources are real and actually say what's claimed — especially with AI output;
4. Ask 'who says, and how would they know?' — consider the source's expertise and incentives;
5. Notice your own wanting-it-to-be-true — the myths we resist checking hardest are the ones we like;
6. Triangulate: multiple independent, credible sources beat one confident assertion.

Field guide: red flags

• Suspiciously round or dramatic numbers with no named source ("90% of people don't know...") — ask where the number comes from;
• No named source at all, or a source that turns out to be another page repeating the same unsourced claim;
• Old news presented as new — check the actual publish date, not just when it crossed your feed;
• Emotionally charged framing designed to make you share before you think;
• A claim that flatters what you already believe — the easiest ones to wave through unchecked;
• A quote or statistic that's suspiciously perfect for the argument being made.

For the specific techniques professional fact-checkers use — lateral reading, reverse image search, spotting satire — see media literacy.

Staying skeptical, not cynical

• Skepticism = proportioning belief to evidence, and being willing to check;
• Cynicism = disbelieving everything, which is just gullibility pointed the other way;
• The goal isn't to trust nothing — it's to trust appropriately, hold beliefs a little loosely, and cheerfully update when you're wrong (which, this whole site suggests, is more often than any of us think).

This isn't just a human problem any more — see how the same failure shows up industrially in AI hallucinations, and where shared false memories come from in how false memories form.