Large language model (LLM)

Definition

A type of AI model trained on very large quantities of language and other information to understand and generate content.

Put simply

It is the underlying model behind many of the conversational and generative AI tools people now use.

What boards and leaders should understand

An LLM is not the same thing as an AI application or agent.

The LLM supplies underlying capability. The finished system may add organisational data, instructions, tools, memory, permissions, integrations and controls.

That distinction becomes important when management says “the model does X”. Often it is actually the whole system around the model producing the result.

Questions to consider

  • When we assess AI performance or risk, are we distinguishing between the underlying model and the complete system built around it?