How does it work?
An AI model turns a piece of text (or an image or audio) into a list of hundreds to thousands of numbers. Together those numbers form a position in a "meaning space". Two sentences that mean the same thing get positions close together, even if they use different words.
Embeddings turn meaning into something a computer can measure: distance.
What is it for?
Embeddings are the invisible building block under many AI applications:
- Semantic search: searching by meaning instead of exact words.
- Vector stores and RAG: the embeddings are stored and made searchable so an AI assistant finds the right passages.
- Recommendations and classification: grouping or suggesting similar items.
Why does it matter?
Without embeddings a system can only match exact words. Embeddings make it possible to work with meaning, which is exactly what an AI assistant needs to understand your content and answer accurately.
Related terms
- Vector store: the database where embeddings are stored and searched.
- Semantic search: searching by meaning, made possible by embeddings.
- RAG: retrieves the right info via embeddings before an AI answers.
