Clicked Gallery

What are Vector Embeddings?

Highlighted from a real engineering doc. Explained by Clicked.

Used in a sentence

The system converted high-dimensional text data into localized vector embeddings to execute semantic search queries.

The reader highlighted one word in the docs. Clicked made the technical term “vector embeddings” easy to understand:

Explained in three depths

●○○

Overview

▾

Vector embeddings are meaning turned into numbers. A model converts text, images or audio into a long list of numbers positioned so that similar meanings land numerically close together, letting computers compare ideas rather than matching words.

●●○

Detail

▾

An embedding model maps each piece of text to a point in a space with hundreds or thousands of dimensions. Distance in that space tracks similarity of meaning, so "car" ends up near "automobile" despite sharing no letters, and a river bank lands far from a money bank when the context differs. That single property powers a lot of modern AI: semantic search that finds documents meaning what you asked rather than containing your words, recommendations, duplicate detection, clustering, and the retrieval step in RAG. Embeddings are stored in vector databases and compared with similarity measures. Two practical caveats: embeddings inherit whatever biases and blind spots their training data had, and general-purpose embeddings can miss the nuances of specialized fields, so legal or medical text often needs domain-tuned models.

●●●

Analogy

▾

A giant flavor map where every dish gets coordinates. Sushi sits near sashimi and far from brownies, while lemon tart lands somewhere between dessert and citrus. Nobody tastes the coordinates, but measuring the distance between two dishes tells you instantly whether they are similar.

AI explanations may contain errors · Not professional advice

Same facts, different vibe — Slang mode 😎

●○○

Overview

▾

Embeddings are how machines understand vibes. Every sentence becomes a big list of numbers, and sentences with the same energy end up as neighbors. The computer never knows what anything means, it just knows what lives near what, which turns out to be most of the job. 😎

●●○

Detail

▾

The trick: meaning becomes location. "Car" and "automobile" share zero letters and land in the same neighborhood. Your search for "how do I undo a commit" finds the doc titled "reverting changes in Git" because they're parked next to each other, not because the words match. That's semantic search, that's recommendations, that's the R in RAG. Fine print: the map was drawn from training data, so it inherited that data's blind spots, and a map drawn from the general internet may fumble your niche legal jargon. Sometimes you need a map drawn by a specialist. 😎

●●●

Analogy

▾

It's Spotify knowing your next song. It has no idea what any track means, it just knows which songs live on the same block. Turns out "lives on the same block" is 90% of understanding.

AI explanations may contain errors · Not professional advice

Formal definition — The same term, explained the usual way

Vector embeddings are dense numerical representations of discrete inputs — tokens, sentences, documents, or other modalities — produced by learned models such that geometric proximity in the embedding space corresponds to semantic similarity. Embeddings underpin semantic retrieval, clustering, recommendation, and retrieval-augmented generation, are typically persisted in approximate-nearest-neighbor indexes, and are compared via metrics such as cosine similarity; representational quality is contingent on the training distribution of the embedding model.

Want Clicked to explain terms like “vector embeddings” directly in your browser — including on PDFs?

Add to Chrome — Free

50 free Explanations · No credit card required

More from the gallery

What is Retrieval-Augmented Generation (RAG)?

Make the AI answer from your documents — retrieval first, generation second, receipts included.

What does an Idempotent API mean?

Why ten clicks charge you once — the property that makes retries safe.

How to actually read a research paper

Skim smart, break the jargon wall, and keep what you learned — one session, done right.

What are “synergies”?

Two merged companies earning more together than apart — the claim, the math, and the donut-shop version.

What is Technical Debt?

Shortcuts and aging systems that bill the future — principal, interest, and repayment.

What is a Zero-Knowledge Proof (ZKP)?

Prove something is true without revealing why — the cryptography behind private verification.