Embedding

An embedding is a way of representing a piece of data — a word, a sentence, a document or an image — as a list of numbers, called a vector, that captures its meaning. The key is that similar items get similar vectors, so they end up "close" to one another in that numeric space: for example, "dog" and "puppy" will be near each other, while "dog" and "aeroplane" will be far apart. AI ✕ Artificial intelligence is a computer system’s ability to perform tasks we associate with human intelligence, such as understanding language, recognizing images or making decisions. More in the glossary → models generate these embeddings, and with them they can compare meanings mathematically, not just exact words. It is the basis of semantic search (finding by sense, not literal match), recommendation systems and techniques like retrieval-augmented generation ✕ RAG connects a language model to an external data source, so it answers based on real, up-to-date information instead of only its memory. More in the glossary → (RAG). Embeddings are often stored in vector databases ✕ A vector database stores embeddings (vectors) and searches them by similarity: given a query vector, it quickly finds the nearest ones — that is, those closest in meaning. More in the glossary → . In essence, they are the bridge between the meaning people understand and the numbers machines process.