What is AI, actually? A plain-English answer

If you’ve been putting off understanding AI because every explanation seems to assume you already understand it — you’re in the right place. Let’s start from zero.

The short answer

The AI everyone is talking about is a computer program that has read an enormous amount of text — books, articles, websites — and learned the patterns in how people write and explain things. When you ask it something, it uses those patterns to write an answer, one word at a time.

That’s it. No feelings, no intentions, no little person inside the machine.

A helpful way to picture it

Imagine someone who has read nearly everything ever written, has a remarkable memory for how things are usually said — but wasn’t actually there for any of it. They never saw a sunset or baked a loaf of bread; they’ve only read millions of descriptions of both.

Ask them anything and they’ll give you a fluent, confident, usually helpful answer. But because they answer from patterns rather than lived experience, they’ll occasionally be confidently wrong — like a well-read friend who sometimes mixes up the details. (That matters so much we wrote a whole guide about it: how to tell when AI is wrong.)

The words you’ll hear, translated

You don’t need to memorise these. Come back to this list whenever a word bothers you.

• AI (artificial intelligence). The umbrella term for computer programs that do things which normally require human thinking — writing, translating, answering questions.
• Chatbot. An AI you talk to by typing (or speaking), the way you’d text a person. ChatGPT, Claude and Gemini are the well-known ones.
• Prompt. A fancy word for the thing you type to the AI. Your question or request. That’s all it means.
• Model. The “brain” behind a chatbot — the program that was trained on all that text. New models come out regularly, like new car models.
• Hallucination. The industry’s odd word for when AI states something false as if it were true. It’s not lying — it has no idea it’s wrong.

Why does it feel so human?

Because it learned from us. It has absorbed the rhythm of human conversation — politeness, humour, encouragement — so talking to it feels like talking to someone. It’s worth remembering that the warmth is a pattern it learned, not a feeling it has. You don’t need to thank it or worry about bothering it (though many of us say “please” anyway — no harm in that).

What it’s genuinely good at — and not

Good at: explaining things at whatever level you ask for, rewriting and summarising text, drafting letters and emails, suggesting ideas, translating, and patiently answering the questions you might feel silly asking a person. Patience is its superpower: it will explain something ten times, ten different ways, and never sigh.

Not good at: knowing what happened five minutes ago, doing precise arithmetic, being right 100% of the time, and knowing you — your health, your finances, your family — beyond what you tell it.

The one idea to take with you

Treat AI like a knowledgeable, tireless, occasionally mistaken helper — not an oracle. Use it to think with you, not for you. If you hold onto that one idea, you’re already using AI more wisely than many people who use it daily.

Ready for the next step? Have your first conversation with an AI — we’ll walk through it together, click by click.