Say hi to Claude

Welcome to bits. No sign-up, no setup and no maths. Let's start by talking to one.

The box below sends your words to Claude, an AI model made by Anthropic. Ask it what it is. The draft is nearly there: add “in one sentence”, then hit Run.

You just talked to a large language model, an LLM. “Large” because it's built from a huge number of internal settings and trained on a huge amount of text. “Language model” because its job is predicting text.

Three words you'll hear everywhere, sorted out once. AI is the umbrella word: software doing things we used to think needed a person. An LLM is the kind of AI that reads and writes text. Claude is one LLM. The chat apps you may have used, like claude.ai or ChatGPT, are a window wrapped around a model like this one. Here you talk to the model with almost nothing in between.

Claude can look at images too. This course sticks to text, because text is where everything else starts.

One honest note about this sandbox. Runs go live to Claude Haiku 5.5, Anthropic's fastest current model (as of October 2026), for free, with an hourly limit that's higher if you sign in. Past the limit you get a saved example, labelled as one, and you can keep going. Answers are kept short on purpose. And each run sends only what's in the box, nothing from before. Why that matters is lesson 6.

Here's a peek at what happened while it answered: the word it picked at each step, and a few it didn't. The numbers are illustrative; the next lesson explains them.