Glossary
What is machine learning?
A way to build software by showing a computer many examples and letting it find the patterns — instead of a human writing out every rule by hand.
In ordinary programming a person writes the rules: if this, then that. That falls apart for messy problems — is this photo a cat? — where nobody can spell out every rule. Machine learning swaps what goes in and what comes out.
| Approach | You give it | It produces |
|---|---|---|
| Traditional programming | Rules + data | Answers |
| Machine learning | Data + answers (examples) | The rules — a model |
ML learns the rules from examples, so it can handle patterns far too fuzzy to write down by hand.
The learned “rules” aren't code — they're millions of parameters tuned during training. Neural networks are today's dominant flavour of ML, and every language model is machine learning run at enormous scale.
Related terms
Appears in these lessons
Learn how
See how AI actually learns
How AI Learns walks through training, data and why more of it helps — the real mechanism, no math degree required.