Skip to content
BENIBIX Learn

Embedding (AI)

Also known as Embeddings, Vector Embedding, Text Embedding

A list of numbers that stands for what a piece of text means.

Think of it like

Think of it like a spot on a map for each idea. Ideas that mean the same live close by.

Example

Emma's help site turns each article into an embedding. 'Reset my password' and 'I can't log in' land close together, though they share no words.

Why it matters

It is how AI search finds text by meaning, not just by matching words.

Definition

An embedding is a list of numbers that stands for the meaning of a piece of data, like a word, a sentence or a picture. The list is often hundreds or a few thousand numbers long. Things that mean close things get lists that sit close together. So a computer can tell how alike two texts are, even when they share no words. Search, grouping and suggestions all run on embeddings.
Each word is a point. Words with close meanings sit close together on the map.

Common mistake

Comparing embeddings made by two different models. Each model has its own map, so their numbers do not line up.

Key takeaways

  1. 1
    Meaning as numbers. One list per item.
  2. 2
    Close means alike. Near lists, near meanings.
  3. 3
    One model, one map. Do not mix them.