AI Knowledge Base
AI Glossary
Structured, sourced and periodically verified reference pages from the AI Knowledge Base.
A software system that uses a model, instructions, and tools to pursue a goal through multiple steps within defined permissions.
AI systems that process or connect two or more types of data, such as text, images, audio, or video.
A model processes inputs and produces outputs; a tool adds a function such as search or calculation; an app combines these in a user-facing product.
A technique that retrieves relevant information from external sources and supplies it to a language model before it generates an answer.
Embeddings are numerical vectors that represent text, images, or other items so their relationships and similarity can be compared mathematically.
An AI hallucination is false or unsupported information generated by a model and presented in fluent, convincing language.
A prompt is the text or set of instructions given to an AI model to define the task it should carry out, including the request itself and any surrounding context, constraints or examples.
A token is a unit of text that an AI model processes when reading or generating language. It may represent a whole word, part of a word, punctuation, or a short group of characters.
The context window is the maximum amount of information, measured in tokens, that an AI model can take into account at one time when producing a response.
Context is the surrounding information an AI model uses to interpret a request, including earlier messages in the conversation, instructions given, and any text or files provided.