AI from Zero · Lesson 1 of 6

What is AI, really?

The honest, jargon-free version of what artificial intelligence actually is, and just as importantly, what it is not.

Beginner9 minute read

Ask ten people what artificial intelligence is and you will get ten answers, most of them from the movies. Let us clear the fog. AI is not a robot brain, it is not alive, and it is not about to take over the world. Underneath the hype, it is something much more down to earth, and once you see it clearly, everything else in this course makes sense.

The simple definition

Here is the version worth remembering: artificial intelligence is getting a computer to do tasks that would normally seem to need human thinking. Recognising a face, understanding a sentence, recommending a film, spotting fraud, playing chess. Tasks where there is no simple, fixed rule to follow, so the computer has to make a smart choice.

That is it. No consciousness, no feelings, no understanding in the way you understand things. Just a machine making useful decisions on tasks that used to require a person.

A quick test

A calculator does maths faster than you, but nobody calls it AI. Why? Because there is one exact rule for adding numbers. AI is for the fuzzy tasks where there is no single rule, like telling a cat from a dog, where the machine has to weigh things up.

Narrow AI vs the sci-fi kind

Almost every AI that exists today is what experts call narrow AI: it does one specific thing, often very well, and nothing else. The AI that recommends your videos cannot drive a car. The one that beats grandmasters at chess cannot write an email.

The all-knowing, human-like machine from films, one system that can do anything a person can, is called general AI, and here is the key fact: it does not exist. Nobody has built one, and no one knows if or when they will. When this course, or the news, says "AI", it almost always means the narrow kind.

What AI is not

It is not conscious. It has no thoughts, feelings, or awareness. It processes numbers.
It is not magic. Every impressive result comes from maths and data, which the rest of this track will demystify.
It is not always right. AI makes confident mistakes, a theme we will return to often.

Try it: spot the AI

Which of these everyday things actually use AI, and which are just ordinary machines following fixed rules? Tap each one to find out.

Interactive

Spot the AI

Tap each item to reveal whether it counts as AI, and why.

Tapped 0 of 8. AI is for fuzzy judgement calls, not fixed rules.

Notice the pattern: AI shows up wherever a fixed rule will not do the job, and the machine has to make a judgement.

Sixty seconds of history

AI has gone through three broad eras. First, machines followed hand-written rules ("if the temperature is above 30, turn on the fan"). Useful, but brittle. Then came learning from data, where instead of writing the rules, we show the machine thousands of examples and let it work out the pattern itself, which is what powers most of today's AI. Most recently, generative AI like ChatGPT can produce brand new text, images, and code. Each era did not replace the last; they stack.

Where we are heading

That middle idea, learning from examples instead of being told the rules, is the single most important concept in modern AI. It is exactly what the next lesson is about.