For Heads-of · Practitioner
Embedding
A numerical representation of text, images or other data in a high-dimensional space, where items with similar meaning sit close together.
- embeddings
- vectors
- semantic search
- representations
Purpose
Embeddings turn raw data into vectors of numbers that a model can compare. Items with similar meaning end up close together, which is what makes semantic search possible.
Example
In classic word embeddings, the vector for king minus man plus woman lands close to queen, showing that the representation captures relationships between meanings, not just spellings.
RAG and search
Embeddings power retrieval-augmented generation (RAG): documents are stored as embeddings and retrieved by similarity to the query. They are not anonymised: an embedding can leak the text it encodes, and a shared vector store can expose one user's content to another. These are the vector and embedding weaknesses that vector store and embedding access controls address.
Related Terms
Data Poisoning
An attack in which an adversary deliberately corrupts data an AI model learns from, to degrade it or plant hidden behaviour. It now reaches beyond training data to fine-tuning sets, retrieval corpora and agent memory, and a poisoned model can pass ordinary evaluation.
Retrieval-Augmented Generation (RAG)
A technique in which a language model retrieves relevant documents or data before generating a response, grounding its output in source material rather than in its training data alone.
Token
A unit of text that a language model processes, typically a word, part of a word or a character. Models work on sequences of tokens, not raw text.
Transformer
A deep learning architecture that uses self-attention to process sequences of data, and the foundation of modern large language models.