Glossary

LAI & Generative Media

Large Language Model

AlsoLLMFoundation Model

An AI system trained on vast text datasets to understand and generate human-like text, powering applications from chatbots to code generation.

A Large Language Model (LLM) is an artificial intelligence system trained on massive text corpora to perform natural language understanding and generation tasks.

Key Characteristics

  • Billions of parameters learned from internet-scale text data
  • Pre-trained on diverse sources, then fine-tuned for specific tasks
  • Exhibits emergent capabilities at scale that smaller models lack
  • Can follow instructions, answer questions, write code, and reason

Notable Examples

  • GPT-4 (OpenAI) - Powers ChatGPT
  • Claude (Anthropic) - Designed with safety focus
  • Gemini (Google) - Multimodal capabilities
  • LLaMA (Meta) - Open-weight research model

Limitations

LLMs can hallucinate facts, lack real-world knowledge after their training cutoff, and struggle with tasks requiring precise reasoning or current information.