What AI is, and what it isn't

If AI feels like a wall of jargon written by people who want you to feel behind, that is not your fault. The technology is usually simpler than the words wrapped around it. You do not need to code and you do not need to understand how it works under the bonnet. You need to know what it is good for.

The honest version

When people say “AI” right now, they almost always mean one thing: a large language model. That is the engine behind ChatGPT, Claude and Gemini.

Here is what it does, without the mystique. It was trained on a vast amount of text, and from that it got startlingly good at predicting what words should come next. That sounds trivial, and it turns out not to be, because predicting the right next words is most of what writing, summarising, translating and answering questions actually is.

The mental model that works

Do not picture a robot brain. Picture a tireless, fast, occasionally overconfident assistant who has read almost everything and will have a go at any writing or reading job you give it.

That framing steers you right nearly every time. Give it words to work with and it shines: first drafts, summaries, tidying up a rushed message, listing the questions a customer might ask. Anything where the input is words and the output is words, and where you still give it the final read.

What it isn’t

It is not a calculator or a source of truth. It has never seen your prices, your policies or your customers, so out of the box it will guess, and its guesses sound plausible. And it is not about to replace you: for an owner it is far more like a keen assistant than a stand-in. It removes the typing and the busywork. The judgement, the relationships and the decisions stay yours.

Keep that one-sentence version in your back pocket: today’s AI is a very good, very fast assistant for words, that you always check. The next lesson is about that last part, because it is the one rule that keeps you safe.