The model doesn't look anything up as it writes. It has learned patterns from its training data and generates a response one piece at a time based on those patterns and your prompt. That's why it can sound fluent on almost any topic and also why it can state something wrong with total confidence.
Modern LLMs are general-purpose. The same model can draft an email, explain a contract, and write code, which is what makes them useful across a business rather than for one narrow task.
Updated July 2026