Last updated: 2026-10-09
Twelve short lessons on how AI tools actually work, when to trust them, how to get good answers, and how to stay safe. No jargon, no coding, and nothing to install.
Self-paced. Honour-based. Not accredited.
If you have never used an AI assistant, or use one without being sure what it is doing, start here. Each lesson explains one idea plainly, gives you a few key terms, and checks your understanding with a single question.
From how a chatbot picks its next word to how to stay in charge of an AI agent.
Tokens, training, search and files: where answers really come from, and what AI is good at.
Why AI makes things up, and a quick way to check the claims that matter.
Clear prompts with a goal, context, format and limits, then a simple loop to improve them.
What not to share, how to judge AI images, and how to spot AI-powered scams.
Choosing tools calmly and keeping control when AI starts doing things for you.
Using AI to learn faster without losing the skills you are trying to build.
Start with the one idea that explains most of how chatbots behave, then work through one lesson at a time.
To go deeper into prompting, retrieval and agents, continue with Beyond the Prompt. For a short practical course you can share with family, try Spot AI Scams and Fake Media. For ready-made prompts, browse the Prompt Library.
Course progress, exam results and certificate details are stored in this browser on this device. They may be removed if you clear browser data.
Ask a chatbot the same question twice and you can get two different answers. Working out why is the quickest way to understand what these tools actually do.
A chatbot's answer can draw on what it learned in training, on a web search, or on files you gave it. Knowing which one you are looking at changes how much you should trust it.
AI is superb at some tasks and unreliable at others that look just as easy. Learning roughly where that line falls saves hours of frustration.
AI tools sometimes state false things with complete confidence. That is not a glitch; it follows from how they work. The fix is a quick, repeatable way to check what matters.
"Write me something" gets you something forgettable. Four short additions turn a vague request into one that gets a useful answer first time.
A clear prompt improves the direction of an answer, not its accuracy. The real skill is a loop: compare what came back with what you wanted, and change one thing.
An AI tool only knows what reaches it. A few simple habits decide what reaches it, and keep your private details, and other people's, out of places they should not be.
AI can now make photos and video that look completely real. The surprising part: looking real was never the right test. Where a picture came from is.
AI has made scam messages better written and cheaper to send, and it can copy voices. The defence that works has not changed: slow down and check through a route you already trust.
Some AI tools now act: they sort files, fill in forms, send messages and run multi-step tasks. That is genuinely useful, and it means your approval needs to sit in the right places.
There are dozens of AI assistants and new ones every week. You don't need the best one; you need a sensible one, set up safely, and a reason before you pay.
The same AI that can teach you something can also do the thinking you meant to practise. Which one happens depends entirely on how you use it.