Start here: AI essentials

Priya uses an AI assistant at work. Some days it saves her an hour. Other days it makes up a rule, or forgets what she said. Your goal: learn why AI acts like this, so you can use it well.

Why this track

Many mistakes with AI come from a wrong picture of it. An assistant is not a search engine or a person. It is an AI model: a very well-read autocomplete that guesses likely text, one small piece at a time.

Where it fits

AI from zero got you started. This track shows the inside. Next, Working with AI puts it to work. We name real products only as examples.

What this track covers

1. What an AI model is — it guesses the next word.
2. Tokens and context windows — what it reads and costs.
3. Inside one answer — why writing is slow.
4. Why answers change: temperature — same question, new answer.
5. When AI makes things up — and why it agrees with you.
6. Prompts that work — write a clear brief.
7. Choosing a model — size, speed and cost.
8. Models that pick from a list — fast, safe choices.
9. Open and closed models — what "open" really means.
10. Search by meaning: embeddings and RAG — find the right text.
11. Prompt, RAG or fine-tune? — pick the right fix.
12. Evals: does your AI work? — test, don't guess.
13. The big picture — the whole track.
14. Cheat sheet & pop quiz

How it works

New words are highlighted. Tap or hover over one to see what it means.

Your progress saves in your browser (the app you use to open websites). The "AI" in each lab is pretend. Nothing you type there is sent anywhere.

Start with what an AI model is.