Glossary

Short definitions of the terms you'll hit while evaluating AI tools.

• LLM (Large Language Model)

  Billions-of-parameters language models trained on vast text corpora. Most generative AI tools run on top of an LLM.

• Prompt

  The input text given to an AI model. Response quality correlates directly with prompt clarity.

• Token

  The smallest unit a language model processes. In English, ~0.75 words per token.

• Fine-tuning

  Process of adapting a pretrained model to a specific task or tone by training further on a narrower dataset.

• RAG (Retrieval-Augmented Generation)

  Architecture where the model retrieves relevant chunks from a knowledge base and adds them to the generation context.

• AI Agent

  AI systems that use tools and APIs to autonomously complete multi-step tasks.

• Embedding

  Technique that converts text/image/other content into meaning-preserving numerical vectors. Foundation of semantic search.

• Hallucination

  AI generating unfounded or fabricated information. Serious use requires verification.