A simple definition
AI is software that learns patterns from very large amounts of examples instead of following only hand-written rules. A spam filter that has seen millions of spam emails learns what spam “looks like”. A chatbot that has read a huge amount of text learns how language, facts and reasoning usually fit together.
The important word is patterns. Modern AI is extremely good at recognising and continuing patterns — in words, images, sound and code. That is why it can write, translate, summarise, draw and program. It is also why it can sound confident while being wrong: a pattern that looks right is not always true.
The three kinds of AI you will meet
- Chat assistants (ChatGPT, Claude, Gemini, Copilot): you type or speak, they answer, write, explain, and analyse files. This is where most people start.
- Generators (Midjourney, DALL·E, Flux, ElevenLabs, video tools): they create new images, voices, music or video from a description.
- Built-in AI features: the “summarise”, “rewrite” or “magic” buttons inside apps you already use — email, documents, phone cameras, scanners, search engines.
What AI is genuinely good at
- Turning rough notes into clean writing, and rewriting text for a different reader.
- Explaining a topic at any level — “explain like I’m 12” or “explain like I’m an engineer”.
- Summarising long documents, meetings and PDFs.
- Brainstorming: ideas, outlines, names, examples, test questions.
- Writing and fixing code, formulas and small scripts.
- Translating and adjusting tone.
What AI is NOT
- Not a search engine by default. Unless it is browsing the web, it answers from what it learned during training, which can be out of date.
- Not always right. It can invent facts, quotes, links and numbers (“hallucinations”). Lesson 6 shows you how to check answers.
- Not a person. It has no feelings or intentions — it produces the most likely helpful response to your input.
- Not a replacement for your judgement. You stay responsible for what you submit, send or publish.
The mindset that works
Treat AI like a very fast, very well-read assistant on their first day at your job. It knows a lot in general but nothing about your specific situation until you tell it. The more context you give, the better the result — and you always review the work before it goes out.