Machine Learning is how AI actually learns. It's the "training" part!
Imagine it like this: You're trying to teach a friend how to tell an apple from an orange, but they've never seen either one. You wouldn't give them rules like "apples are round and red" — that doesn't always work, some apples are green!
Instead, you would show them hundreds of pictures: "This is an apple. This is an orange. This is an apple…" After enough pictures, your friend starts noticing the patterns on their own — the shape, the texture, the little stem — and can guess correctly even on a fruit they've never seen!
That's exactly what Machine Learning does, except the "friend" is a computer program, and instead of fruit photos it can learn from:
So? We don't give the computer the exact rules — we give it examples, and it finds the rules on its own. The more good examples it sees, the smarter it becomes.