Lesson 2: Prompts & Tokens

What You'll Learn

What a prompt is and why the way you ask matters so much
What a token is — the small chunk of text an AI reads and writes in
What "context" means and a few simple habits for talking to an AI

What Is a Prompt?

A prompt is simply the instruction or question you give to an AI. When you type "Write me a short poem about autumn" into ChatGPT or Claude, that whole sentence is your prompt. The AI reads it and responds.

Here is the key idea: the AI can only work with what you give it. It cannot read your mind. A vague prompt usually gets a vague answer, and a clear, detailed prompt usually gets a much better one. Think of it like ordering at a restaurant. If you say "I'll have food," you might get anything. If you say "I'll have the tomato soup, no croutons, with a side of bread," you get exactly what you wanted.

So instead of "Write about dogs," try "Write a friendly, 100-word paragraph for a children's book about why dogs make good pets." The second prompt tells the AI the tone (friendly), the length (100 words), the audience (children), and the angle (why dogs make good pets). More guidance in, better results out. The skill of writing good prompts even has a nickname: prompting, and it gets easier with practice.

What Is a Token?

When you read, you naturally see whole words. An AI works a little differently. It breaks text into small chunks called tokens. A token is roughly a word-piece — sometimes a whole short word like "cat," and sometimes just part of a longer word. For example, the word "unbelievable" might be split into a few tokens like "un," "believ," and "able."

A rough rule of thumb: one token is about three-quarters of a word in English. So a paragraph of 100 words is very roughly 130 tokens. You do not need to count tokens yourself, but it helps to know the AI is reading and writing in these little pieces rather than whole sentences at once. This is also why, in the last lesson, we said the model predicts the "next chunk" — that chunk is a token.

Tokens matter for one practical reason: AI tools have limits on how many tokens they can handle at once. Very long documents may need to be shortened or broken into parts.

Context: What the AI Can "See"

Context is everything the AI is paying attention to right now — your current prompt, plus the earlier messages in your conversation. It is like the AI's short-term memory for your chat.

This is why follow-up questions work. If you ask "Who painted the Mona Lisa?" and then ask "When was he born?", the AI understands "he" means Leonardo da Vinci, because your first message is still in the context. But context has limits. In a very long conversation, the earliest messages can drop out of view, and the AI may "forget" something you said much earlier. If that happens, just remind it.

A few simple habits make talking to an AI easier: be specific, give an example of what you want when you can, and feel free to ask it to try again or adjust ("make it shorter," "make it more formal"). You are having a conversation, not casting a spell — you can always refine.

Key Takeaways

A prompt is the instruction or question you give an AI; clearer, more detailed prompts get better answers.
A token is the small chunk of text — roughly a word-piece — that the AI reads and writes in.
Context is what the AI is currently paying attention to, including earlier messages, and it has limits.