RolloAcademy
Core conceptUnder the hood

Machine learning

The main way modern AI is built: the system learns patterns from many examples rather than being handed explicit rules.

Machine learning: learn a pattern from labelled examples, then judge new cases Labelled example emails feed a box that learns the pattern; a new email then gets sorted as spam or not. LABELLED EXAMPLES Email → not spam Email → spam Email → not spam Learns the pattern New email Spam?

Machine learning is the engine under most of today's AI. You show the system thousands of examples, say, emails marked "spam" or "not spam", and it works out the patterns for itself, then applies them to things it has never seen.

It is a part of artificial intelligence, not a separate thing. Product recommendations, fraud alerts, and photo tagging are all everyday machine learning at work.

Often confused with: AI in general. Machine learning is one way to build AI, the most common one today, not another word for the whole field.

See it in a lesson: What is AI?

Related terms

Artificial intelligenceGenerative AI

Further reading

← All terms