What AI really is (beyond the hype)

Why this matters to you now

You already use AI many times a day — when a search ranks results, a phone unlocks by your face, a map predicts traffic, a feed decides what you see next, or a classmate uses a chatbot for homework. AI is not coming; it is here, woven into ordinary tools. That makes understanding it a basic life skill, like knowing how money or the internet works. This course gives you that understanding honestly: what AI can do, what it cannot, and how to use it well without being fooled or getting into trouble.

Two ways to make a computer do something

For most of computing history, humans wrote rules: "if the total is over 1000, apply a discount." A programmer imagines every case and writes instructions for it. This is ordinary programming, and it is exact and predictable.

Some problems, though, are almost impossible to write rules for. Write the rules that tell a cat from a dog in a photo. Pointy ears? Foxes have those. Whiskers? So do seals. You would never finish. Artificial intelligence takes a different route: instead of rules, you give the computer thousands of examples — photos labelled "cat" or "dog" — and it works out the patterns itself. That approach is called machine learning, and it powers almost everything called "AI" today.

So the one-line difference to remember: ordinary software follows rules a person wrote; machine learning finds patterns in data. This single idea explains most of AI's strengths and every one of its weaknesses.

Narrow AI vs the AI of films

In films, AI is a conscious machine that thinks, feels and wants — artificial general intelligence (AGI), able to do anything a human mind can. That does not exist. What exists today is narrow AI: systems that are very good at one specific task and useless outside it. The AI that beats grandmasters at chess cannot make you a cup of tea or understand a joke. A chatbot that writes fluent essays cannot reliably add up a column of numbers.

This matters because the word "intelligence" misleads. These systems are powerful pattern-matchers, not minds. They do not understand what they produce the way you understand this sentence. Keeping that in mind is the difference between using AI wisely and trusting it blindly.

A quick map of the words you'll hear

• Artificial Intelligence (AI): the broad field of making computers do tasks that seem to need human intelligence.
• Machine Learning (ML): the main method today — learning patterns from data instead of following written rules.
• Deep learning / neural networks: a powerful kind of machine learning, loosely inspired by brain cells, behind image recognition and modern chatbots.
• Generative AI: AI that *creates* new things — text, images, audio, code. ChatGPT, Gemini and image generators are generative AI.
• Large Language Model (LLM): the kind of AI behind text chatbots, trained on huge amounts of writing to predict text. Lesson 3 opens this up.

You do not need to memorise these. You will meet each one in context; this is just the map.

What AI is genuinely good at — and not

Good at: spotting patterns in huge amounts of data, recognising images and speech, translating languages, drafting and summarising text, suggesting ideas, and doing repetitive language or image work quickly.

Bad at: being reliably correct about facts, understanding meaning and context the way people do, reasoning through genuinely new problems, and anything requiring real-world judgement, ethics or care. It also cannot tell you when it is wrong — it produces mistakes with the same confidence as correct answers.

Try it

• List every AI you used yesterday. For each, ask: is it finding patterns in data, or following simple rules? Some (a calculator) are just rules; most (recommendations, voice, cameras) are learned.
• Find one recent news headline about AI. Does it describe a real narrow-AI tool, or imply science-fiction AGI? Notice how often the two get blurred.
• Ask people at home what they think AI is. Compare their answers with the rules-vs-examples idea in this lesson.

Discussion question

If AI learns from examples made by humans, what might it learn that we would not want it to? Hold that thought — it is the subject of lesson 5.