Neural networks
A neural network is a computer model made of layers of simple connected units that learn to recognize patterns by adjusting the strength of their connections.
The name comes from a rough resemblance to brain cells passing signals to each other. In a computer, each unit is a tiny calculation: it takes numbers in, weighs them, adds them up, and passes a result on. Alone, a unit is trivial. Wired together in layers, thousands or billions of them can recognize faces or write sentences.
Learning means changing the weights on the connections. At the start they are random and the network's output is garbage. Training shows it examples, measures the error, and nudges the weights to reduce it. Slowly, the network organizes itself so that useful features are detected at each stage.
Neural networks are not brains. They do not grow new connections, they do not experience anything, and the brain analogy stops early. They are best thought of as extremely flexible pattern-fitting machines. Deep learning, transformers, and large language models are all types of neural network.
When a music app identifies a song from a few seconds of your humming, a neural network is matching patterns in the sound to millions of songs it learned.