What Is an Embedding?
Also called vector embedding
Short Answer
An embedding is a list of numbers, a vector, that represents the meaning of a piece of text, an image or another item. An embedding model places similar meanings close together, so "car" and "automobile" get nearby vectors. Embeddings power semantic search, recommendations and retrieval-augmented generation.
- Taught in lesson 98
- In 1 real job ad
Why It Matters
Embeddings let computers compare meaning, which powers semantic search, recommendations and the retrieval step of RAG assistants. Anyone building AI search or document assistants works with them, and understanding them explains why some searches find the right passage and others miss it.
How It Works
Embeddings typically have hundreds to a few thousand dimensions. Closeness is measured with cosine similarity or distance: the closer two vectors, the more similar the meanings.
In a RAG system, every document chunk is embedded once and stored in a vector database. Each question is embedded the same way, and the nearest chunks are retrieved and given to the language model.
An embedding model is trained so that related items end up close together in a space with hundreds or thousands of dimensions. You cannot read meaning from a single number, but distances between vectors carry it. Embeddings work best for the kind of text the model was trained on; specialised fields such as medicine or law sometimes need domain-specific embedding models to capture their vocabulary well.
Embedding in Practice
Everyday Example
A search for "how to reset my password" finds a help article titled "Recover access to your account", because their embeddings are close even with no shared words.
Industry Example: Pharma
Embedding past deviation reports lets a quality team find similar earlier incidents, even when they were described with different words, to speed up root cause analysis.
Where You Meet It
- Semantic search over documents.
- The indexing step of RAG systems.
- Recommendation and duplicate-detection features.
- Vector databases.
Key Ideas Around It
- Vector
- A list of numbers representing an item.
- Dimension
- One of the numbers in the vector.
- Cosine similarity
- A measure of how close two vectors are in direction.
- Embedding model
- The model that produces the vectors.
A Common Mix-Up
Embeddings from different models cannot be compared with each other. Every item must be embedded with the same model.
Check YourselfWhy do "car" and "automobile" get similar embeddings?
Because the embedding model learned that they are used in similar contexts, so it places them close together.
Seen in Real Job Ads
Lines quoted word for word from live job postings.
Pfizer
Senior Manager, LLMOps & AI Integration Engineer
Hands‑on with GenAI/LLM ecosystems (e.g., tool/function calling, structured outputs, prompt orchestration) and RAG components (embeddings, retrievers, rerankers, vector stores).
Learn It on ai2hired
S9 · #98Embeddings
Coming in Season 9, Embeddings, RAG and Search. Every ai2hired lesson is free on YouTube.
Questions People Ask
Is an embedding the same as a language model?
No. An embedding model turns text into a vector; a language model generates text. RAG systems use both.
Can embeddings work across languages?
Multilingual embedding models place the same meaning in different languages close together, enabling cross-language search.
Related Terms
Learn AI From Zero to Hired
Free lessons on YouTube, each one built from what real AI job ads ask for.