Why models make things up

Because a model writes the most plausible continuation, it can produce fluent text that is false: a made-up citation, a wrong date, an invented quote. It isn't looking facts up unless it's given a search tool or documents to read. Good habits: ask for sources and check them, give the model the material to work from, and treat confident wording as style, not proof.

Why it matters: Understanding this is what turns AI from a mystery into a tool you can use well.

Builds on: Next-token prediction, Training versus using a model

Learn it, then prove it

Sign in (free) to take this step's mastery check and save your journey.