Decoder. plain-English AI glossary

Neural network

● Core

The layered math structure underneath modern AI — loosely inspired by brain neurons, but really just numbers flowing through connections.

Think of it like

A relay race where each layer passes a refined version of the signal to the next.

Example

Every LLM, image generator, and speech model is some flavor of neural network.

How it actually works

Input becomes numbers, those numbers pass through layers of weighted sums and simple non-linear functions, and out comes a prediction. Training compares the prediction to the truth and adjusts the weights to shrink the error. “Deep” just means many layers. The brain analogy is mostly marketing — it’s linear algebra, not biology.

For product teams

The umbrella technology; you rarely touch it directly, but it’s what “AI model” refers to.

For engineers

Stacked layers of linear transforms plus non-linearities, trained by backpropagation.

Weights — Its connections are the weights.
LLM — The transformer is one influential design of one.

Read anything AI without the jargon

Look up any term in plain English, or save terms as you read with the free Chrome extension.

Open Decoder Add to Chrome