Foundations
What Is a Neural Network? The Basics Behind Every AI Model
Every modern AI model — LLMs included — is a neural network under the hood. Here's what that actually means, without the math.
Loosely inspired by neurons
A neural network is a system of simple, interconnected units ("neurons") organized in layers, loosely inspired by how brain neurons connect. Each unit takes in numbers, applies a simple transformation, and passes the result forward. No individual unit is smart — the capability emerges from millions or billions of them working together.
What "training" actually adjusts
Each connection between units has a weight — a number that determines how strongly one unit's output influences the next. Training is the process of gradually adjusting all those weights (see What Is Gradient Descent?) so the network's output gets closer to what it should be, across a huge number of examples.
From basic networks to LLMs
A large language model is a neural network with a specific architecture (the transformer, built around attention) and a specific training objective (predict the next word). The underlying building block — layers of connected units with adjustable weights — is the same idea used across nearly all of modern AI, from image recognition to speech.
Frequently Asked Questions
What is a neural network in simple terms?
A system of simple, connected units organized in layers that collectively learn to transform an input into a useful output, by adjusting the strength of the connections between units during training.
Are LLMs a type of neural network?
Yes — an LLM is a neural network using the transformer architecture, trained specifically to predict the next piece of text.