Spotting hallucinations before they cost you.

Every AI you've used has confidently lied to you at some point. It's not a bug — it's how the technology works. You can't stop hallucinations, so the skill is catching them fast.

You'll watch an AI invent something believable, learn why it happens, then build a prompt that's much harder to fool. About 9 minutes.

01 · Catch one in the wild

Here's a real-shaped answer an AI gave to "What does the research say about the four-day work week?" Read it like you'd read it at work:

Which part should you not repeat in a work email without checking?

02 · Why AI invents things

A language model predicts the most plausible next words from patterns in its training data. There's no fact being looked up. Ask "what year did X happen?" and it isn't checking a record — it's predicting what year would sound right. A precise-looking citation is highly "plausible," and plausible isn't true.

So why did the AI invent that Cambridge study?

03 · The four high-risk patterns

Hallucinations aren't random — they cluster. These four are where to raise your guard:

You're skimming an AI draft for a client email. Which line should make you stop and verify?

04 · Build a hallucination-resistant prompt

You can cut the odds before they happen. Start with a risky factual prompt, then add each guardrail and watch your protection climb:

The rule to keep: chat AIs for writing, research-grounded AIs for facts. Give the model permission to say "I don't know," ask for sources and actually check them, and raise your guard on numbers, names, anything current, and technical syntax.