What this course is about

AI chatbots like ChatGPT, Claude and Gemini are programs you talk to in plain language. They can write, explain, summarise and answer questions. This course is about using them well.

They’re genuinely useful, and also wrong in ways that are easy to miss: they can invent a fact, a quote or a reference and state it with total confidence. Using them well means knowing why that happens, how to ask clearly, how to check what you get back, and what you should never paste into one.

This course is for someone who has never seriously used one and doesn’t write code. Eight short notes, readable in an evening, with no maths, no code, and every term explained where it first appears.

If you already program and want to build things with AI, skip this and start at the AI Engineer course.

What you’ll be able to do

By the end you will be able to:

  • explain in plain words what an AI chatbot is and why it makes the mistakes it does;
  • pick the right kind of AI tool for a task;
  • ask clearly, so you get useful answers first time;
  • check an answer properly, and know which answers need checking most;
  • keep private and work information safe;
  • decide where AI helps you and where it gets in the way of learning.

Where to start

Read notes 01–06 in order; each builds on the one before. Notes 07 and 08 can be read any time. If you use these tools at work, read 07 early.

  1. What This Thing Actually Is: what AI, machine learning and a “model” mean, and what today’s AI can and can’t do.
  2. How LLMs Work, Plainly: these chatbots predict the next word, one at a time, and that one fact explains why they miscount, invent references, and answer differently each time.
  3. Choosing a Tool: chat, search, research, document, image and voice tools; free or paid; and three questions that pick the right one.
  4. Talking to a Model: asking well is like briefing a new colleague: the task, the background, the format and who it’s for. The six techniques that matter, and the myths that don’t.
  5. Context and Long Chats: give it the document instead of asking from memory, why long chats get worse, and when to start a new one.
  6. Verifying What It Tells You: the most important note here. Why sounding confident means nothing, what to check hardest, and why asking “are you sure?” isn’t a check.
  7. Privacy and What Not to Share: what happens to what you type, removing private details first, and rules for work and client information.
  8. Living With It: where it really helps, the risk of losing skills you stop practising, when to say you used it, and where to go next.

Practice

Reading isn’t enough. Use it for something real: a document you actually need summarised, a letter you need to write, a topic you want explained. Do that for two weeks, then re-read note 06. It lands differently once a chatbot has confidently misled you.

Key ideas to remember

  1. It guesses the next word; it doesn’t look things up. Every quirk follows from that. → 02
  2. Give it the source material. Attaching your document beats asking it from memory, every time. → 05
  3. Ask yourself: “would I notice if this were wrong?” If not, check it somewhere outside the chat. → 06
  • AI & ML: how AI works, for people who want to build with it.
  • AI Automation: making AI tasks run automatically, on a schedule or when something happens.
  • The home page: the map of every subject.

About this course

  • Planned: a course for technical people who don’t yet build with AI: coding assistants, AI-assisted debugging and code review, what to let an AI agent change, and reviewing code you didn’t write. It would sit between this course and the AI Engineer course.
↑ Back to Kingsley's Knowledge Base