What AI actually is and how it thinks

What we mean by "AI" today

When someone says AI today, they almost always mean a large language model - an LLM. ChatGPT, Claude and Gemini all belong here. It is a tool that can read and write in human language: it answers questions, summarises, translates and writes.

It is not a robot from the movies, nor a consciousness with opinions of its own. It is a program that learned from an enormous amount of text from the internet and from books.

How it "thinks" (a simple but honest picture)

The simplest honest explanation: AI predicts the next word. It reads what you wrote and guesses which word makes the most sense next. Then the next one. And the next. That is how a whole answer comes out.

It sounds almost silly, but across a vast amount of text it turns into something surprisingly clever - something that can answer, advise and write.

It is like hearing "in the morning I have my coffee with ___" and your brain filling in "milk" by itself. AI does exactly that, just at enormous scale and on every topic there is.

Why it comes across as so clever

Because it has seen billions of sentences, it can imitate an expert, a friend or a lawyer. It can talk about cooking and about the law. But careful - it does not look into the truth, only into patterns in text. That is why it will occasionally say something completely wrong with total confidence. There is a whole lesson on that later in the course.

What that means for you

Treat AI as an extremely well-read and fast, but occasionally unreliable assistant. It is brilliant for a first draft, for ideas, for summaries. It is not a source of final truth.

And above all: the better you brief it - context, an example, exactly what you want - the better it answers. You will pick that skill up in the prompting course. For now it is enough that you know what you are dealing with.