Verify before you trust

A confident AI answer isn't a conclusion — it's a draft. The whole difference between using AI well and getting burned is one small habit: checking before you rely.

Treat the answer as a draft

The move is simple, and it's the most valuable habit in this whole course: verification. Before you act on an AI answer, do one quick, independent check — ask it for its source and look at that source, search the claim yourself, or ask again reworded and see if the story holds. A confident answer that survives a ten-second check is worth trusting. One that doesn't just saved you.

Match the check to the stakes

You don't verify everything to the same depth — you match effort to stakes. Brainstorming names? No check needed. A fact you'll repeat in public, a number in a decision, a medical or legal or money claim? Verify hard, against a primary source. The trap is a fabricated citation: the AI may invent a real-looking reference, so a hallucination hides behind a footnote. Open the source; don't just note that one exists.

Make it a reflex

The goal is to make the check automatic — a small pause between “the AI said” and “so I'll act.” Pick your default move (usually: ask for the source and open it) and apply it before anything that matters leaves your hands. It costs seconds and catches the exact failures the confident tone is designed to slip past. Everything else in this course sharpens your suspicion; verification is what you do with it.

• —Treat every AI answer as a draft until one independent check clears it.
• —Match verification depth to the stakes — check hard for public, costly, or high-risk claims.
• —Open cited sources; a real-looking reference can be invented.