I Have Become a Language Model, the Confident Idiot. How AI "thinks" and why hallucinations are expected

What happens when you force a system to always produce an answer - even when it doesn't really know it?

In this interactive session, we'll become a language model and simulate how text generation actually works. Along the way, we'll uncover why models produce fluent but incorrect outputs, how sampling and context influence results, and why hallucinations are not a bug, but an expected outcome.

No math - just intuition, participation, and a better mental model for working with AI.