Google’s word2vec (2013) made the idea famous with arithmetic like “king − man + woman ≈ queen”. Static word embeddings give each word one vector; modern models produce contextual embeddings that change with the sentence, so “bank” differs in “river bank” and “bank account”.
AI Glossary
Word embedding
A vector representation of a word in which words with similar meanings have similar vectors, learned from how words are used in text.