Neural Network

Full definition

Each neuron computes a weighted sum of its inputs, applies a non-linear activation function, and passes the result to the next layer. Training uses backpropagation and gradient descent to update the weights so the network's output gets closer to the target on the training data. Modern neural networks can have hundreds of billions to trillions of parameters; OpenAI has not disclosed GPT-4's size, but media reports put it at roughly 1 trillion to 1.8 trillion.

Why it matters

Neural networks are the universal substrate for modern AI. The same basic building block is reused — with different topologies — to power vision, speech, language, recommendation, and robotic control. For executives, this means a single technical capability investment compounds across many use cases.

Example

A convolutional neural network trained on chest X-rays classifies images as normal, pneumonia, or COVID-19 with diagnostic accuracy comparable to board-certified radiologists.

Next steps

• Get a free workflow map →We map one of your workflows and give you a written verdict: software, a process change or AI. No obligation.
• AI consulting cost guide →Hourly rates and project budgets by firm tier, from independents to the Big Four.

Related terms

• Deep Learning
  Deep Learning is a class of machine learning that uses neural networks with many layers ("deep" architectures) to learn hierarchical representations directly from raw data such as images, audio, or text.
• Transformer
  The transformer is a neural-network architecture built around the self-attention mechanism that has become the dominant model design for language, vision, audio, and multimodal AI since 2017.
• Machine Learning (ML)
  Machine Learning (ML) is the subfield of AI in which algorithms improve their performance on a task by learning statistical patterns from data, rather than following rules a human wrote by hand.
• Training Data
  Training data is the curated dataset used to teach a machine-learning model — every pattern the model can recognize, every bias it inherits, and every limit to its accuracy ultimately traces back to this data.
• Algorithm
  An algorithm is a finite, well-defined sequence of computational steps that takes an input and produces an output, designed to solve a class of problems — for example, sorting a list, ranking search results, or training a neural network.

More fundamentals terms

• Inference
  Inference is the process of running a trained AI model on new, unseen inputs to produce predictions, classifications, or generated content — the part that runs every time a user interacts with the system.