A large language model, abbreviated LLM, is a neural network trained on very large quantities of text to predict and generate language. ChatGPT, Gemini and Claude are all built on LLMs.
LLMs use the transformer architecture and learn statistical patterns across billions of text examples. They generate output by predicting likely continuations rather than retrieving stored facts, which explains both their fluency and their tendency to fabricate. Training data has a cutoff date, so knowledge of recent events requires retrieval.
LLMs now sit between users and websites. They absorb informational queries, summarise content that previously required a visit, and decide which sources to cite when retrieving live information. Understanding how they select and use sources has become part of understanding search visibility.
Write content that retrieval systems can use confidently: clear structure, explicit facts, unambiguous statements and corroboration from elsewhere. Decide deliberately which AI crawlers to allow in robots.txt, balancing training use against the visibility that comes from being retrievable.
Want to learn more? Explore our SEO Glossary or check out our expert guides on SEO strategies.