Phase 1 · Building with LLMs · Prompting
What an LLM actually does
Next-token prediction, and why that produces useful behavior.
Why it matters
You can send a request and read the reply. To build something reliable on that call, you need a working picture of what happens in between. Most prompting advice in Phase 1, and most of the ways LLM apps fail, follow from one fact: the model writes its reply by predicting the next token, over and over. Once that clicks, you'll know when to trust a model, when to check it, and when to hand the job to ordinary code.
What you'll be able to do
- Describe how a model writes a reply one token at a time
- Explain where a model's knowledge comes from, and why it can sound sure and still be wrong
- Spot tasks that need checking or plain code instead: counting, recent facts, exact math
- Simulate the generation loop in Python with a toy probability table
Before this lesson
Next in this phase: Tokens and the context window