Machines that learn from examples

Could you write the rules?

You can tell a cat from a dog in a fraction of a second. But try writing down the rules you use. Pointy ears? Some dogs have them. Size? Some cats are huge. Whiskers? Both. Every rule you write breaks on some animal.

For decades, that’s how people tried to make computers smart: by writing rules. It works for chess openings and tax forms, but it falls apart for anything messy — faces, voices, language, the real world.

Machine learning flips the approach. Instead of writing the rules, you show the computer lots of examples with the right answers, and it works out the rules itself.