Decoder. plain-English AI glossary

Embeddings

● Core

Turning words into coordinates, so a computer can measure how close in meaning two things are.

Think of it like

A map where “dog” and “puppy” are next-door neighbors and “dog” and “tax form” are on opposite coasts.

Example

Search that finds “car” when you typed “automobile” is matching on embeddings, not on spelling.

How it actually works

Each piece of text becomes a long list of numbers — a vector. “Similar” means the vectors point in nearly the same direction (cosine similarity). The entire trick of semantic search lives in this one idea.

For product teams

The tech behind 'find similar' — powers semantic search, dedup, and recommendations.

For engineers

Dense vector representations where cosine distance approximates semantic similarity.

RAG — The engine underneath RAG.
Tokens — Not the same as tokens, though both are “text as numbers.”

Read anything AI without the jargon

Look up any term in plain English, or save terms as you read with the free Chrome extension.

Open Decoder Add to Chrome