📈 What is machine learning?

Traditional programming is you writing rules: 'if this, then that'. Machine learning flips it — you give the computer lots of examples and the right answers, and it figures out the rules itself.

Think of teaching a child to recognise a dog. You don't hand them a list of rules about fur and snouts. You show them dogs, and they build the pattern. Machine learning is the same: show a computer ten thousand pictures labelled 'dog' and 'not dog', and it learns to tell them apart.

There are a few main flavours. 'Supervised learning' means you have the answers (these emails are spam, these aren't). 'Unsupervised learning' means the computer finds structure on its own (grouping customers with similar habits). 'Reinforcement learning' means it learns by trial, error and reward (playing a game and getting points).

Everything impressive in AI right now — chatbots, image generators, recommendations — is some form of machine learning. It's the reason computers can now do things nobody could write rules for.