Large Language Model
A neural network trained on large volumes of text that can generate, summarise, translate and reason about language.
In detail
A large language model (LLM) is a type of foundation model trained on billions of tokens of text using a transformer architecture. During training the model learns statistical patterns across language, code, structured data and other text. At inference time it generates a continuation of any input prompt by predicting the most likely next token. Modern LLMs have emergent capabilities including instruction following, multi-step reasoning, code generation and structured output production that are not explicitly trained but arise from scale. Examples include Anthropic's Claude family, OpenAI's GPT series, Google's Gemini and Meta's Llama models.
Why it matters for Australian business
LLMs are the engine behind most AI tools Australian businesses are using today: ChatGPT, Microsoft Copilot, Claude, Gemini and the AI assistants built into Xero, HubSpot and similar platforms. Understanding that these are probabilistic text-prediction systems, not databases or reasoning engines, shapes how you use them safely. Key implications: they hallucinate, they need grounding with real data (RAG), they cannot be fully trusted on sensitive decisions without human review, and the data you send to them is processed by the provider's infrastructure.