Large Language Models (LLMs) are a class of artificial intelligence (AI) systems that consist of very large and diverse collections of text (and often code, images, and other data) that recognize patterns in language and generate human-like responses. LLMs are designed as general-purpose models capable of performing a wide range of tasks, such as summarizing, drafting, translating, reasoning, and answering questions across many domains.

LLMs do not “know” facts in a human sense; rather, they predict responses based on statistical patterns learned during training. Examples of LLMs include ChatGPT (OpenAI), Gemini (Google), LLaMA (Meta AI), Bard (Google AI), and Claude (Anthropic).

Depending on how they are deployed, LLMs may be further refined through techniques such as fine-tuning, feedback loops, or reinforcement learning, often with governance controls that limit how user interactions are retained or used.

In the context of the learn-and-work ecosystem, LLMs are useful for their breadth, flexibility, and ability to provide information and support exploration by users. Examples of use include career navigation, content generation, and knowledge synthesis. Their general-purpose design results in varying levels of accuracy, consistency, and explainability. For this reason, human oversight and organizational guardrails are needed to work effectively with LLMs.

See Glossary: Closed-Circuit AI Models (Domain-Specific or Enterprise AI Models) | Learn & Work Ecosystem Library

See Glossary: Opacity (in Artificial Intelligence) | Opaque AI | Learn & Work Ecosystem Library

See Topic Brief: AI Architectures in the Workplace: Large Language Models & Closed-Circuit AI Systems | Learn & Work Ecosystem Library