What is a neural network?
A neural network is an AI model made of many simple connected units, loosely inspired by neurons in the brain. Nobody writes its rules: it learns patterns from examples by gradually adjusting the strength of its connections.
Updated September 24, 2026
A neural network is the kind of AI model behind most of today's breakthroughs, from chatbots to face recognition. It's built from many simple units, often called artificial neurons, wired together in layers. Instead of following hand-written rules, the network learns from examples by adjusting how strongly its units are connected.
How it works
- Layers. Data enters through an input layer, is transformed step by step in hidden layers, and comes out as a prediction from the output layer.
- Weights. Every connection has a weight, a number that sets how much one unit influences the next. IBM compares weights to dials.
- Learning. The network makes a guess, measures how wrong it was, and sends that error backward through the layers, nudging every weight to do better next time. This process, called backpropagation, repeats millions of times.
Main types
- Convolutional networks (CNNs) scan images with small filters, great for vision.
- Recurrent networks (RNNs) loop information back, suited to sequences like speech.
- Transformers use "attention" to weigh which parts of the input matter most. They power large language models.
Networks with many hidden layers are called deep learning, made practical by fast GPUs.
Where they're used
Self-driving cars and face recognition, speech recognition and translation on your phone, chatbots, and the recommendations in shopping apps.
A long history
The idea dates to the 1940s, when researchers Warren McCulloch and Walter Pitts proposed a mathematical model of neurons. Frank Rosenblatt's Perceptron in 1957 was the first network that could be trained. Interest faded, then returned in the 1980s and exploded in the 2010s.
In 2024, John Hopfield and Geoffrey Hinton won the Nobel Prize in Physics for foundational discoveries that made machine learning with neural networks possible.
Quick questions
Is a neural network like a brain?
Only loosely. The design was inspired by the brain: nodes stand in for neurons, and connections that grow stronger or weaker are compared to synapses. But it's math running on chips, not biology.
What is deep learning?
Machine learning that uses neural networks with many hidden layers, often dozens. Graphics chips (GPUs) made training networks that deep practical.
What is the difference between a neural network and machine learning?
Neural networks are one method within machine learning. Traditional methods often need people to pick out useful features, while deep neural networks can learn them from raw data.
Who won the Nobel Prize for neural networks?
John Hopfield and Geoffrey Hinton won the 2024 Nobel Prize in Physics for foundational work that made machine learning with artificial neural networks possible.
Sources: MIT News: Explained: Neural networks, IBM: What is a neural network?, AWS: What is a neural network?, Nobel Prize in Physics 2024: press release