How AI learns
The single most important idea in modern AI, with zero maths: how a computer gets smart by seeing examples instead of being told the rules.
Old-fashioned software follows rules a human wrote. But how do you write a rule for "is this a cat?" You cannot, not really. There are too many cats, too many angles, too many exceptions. So modern AI does something cleverer: instead of being told the rules, it is shown thousands of examples and works out the pattern for itself. That process is called machine learning, and it is the engine behind almost every AI you have heard of.
Rules vs examples
Picture teaching a child what a dog is. You do not hand them a rulebook ("four legs, fur, wet nose, barks"). You point at dogs. "That's a dog. That's a dog too. That one's a cat." After enough examples, they just get it, even for breeds they have never seen.
Machine learning works the same way. You feed the computer a big pile of labelled examples, the training data, and it gradually adjusts itself until it can label new ones correctly. Nobody ever writes the rule for "dog". The machine discovers it from the examples.
What is "a model"?
You will hear the word model constantly. It just means the thing that comes out of training: the pattern the computer learned, saved so it can be used on new data. Training a model is the learning phase; using the model to make a prediction is called inference. When people say "an AI model", they mean this learned pattern, nothing more mysterious.
Three ways to learn
There are three broad styles of machine learning. You do not need the details yet, just the flavour of each.
Try it: more examples, smarter machine
The single biggest factor in how well an AI learns is how much good data it sees. Drag the slider to give a model more training examples and watch how accurate it gets.