Large Language Model (LLM)

An AI model trained on vast amounts of text data, capable of understanding and generating human-like text.

How it works

LLMs are built on the Transformer architecture and trained on massive datasets comprising books, articles, and websites to learn the statistical patterns of human language. By predicting the next word (or token) in a sequence billions of times, the model develops deep contextual understanding of grammar, facts, and reasoning. After pre-training, they are fine-tuned with techniques like RLHF to follow instructions helpfully and safely.

Why it matters

Large Language Models have revolutionised human-computer interaction. They serve as the foundation for chatbots, code assistants, document summarisation, and automated reasoning tools. Their ability to process and generate natural language at scale is driving the current AI boom, making technology more accessible and enabling new classes of software that can understand unstructured text — which represents the vast majority of human knowledge.