← Glossary · Models

BERT

Model

Fact-checked May 20, 2026

Also called: Bidirectional Encoder Representations from Transformers

BERT stands for Bidirectional Encoder Representations from Transformers. It's a popular natural language processing model developed by Google.

BERT is a powerful language model released by Google in 2018. It revolutionized how computers understand human language by being one of the first models to process words in relation to all the other words in a sentence, both from left-to-right and right-to-left, which is what 'bidirectional' means. Before BERT, most models would only look at words in one direction.

This bidirectional approach allows BERT to understand context much better than previous models. For example, in the sentence 'The bank of the river' versus 'The bank where I keep my money,' BERT can distinguish the meaning of 'bank' because it sees the whole sentence at once. This capability made it very successful for tasks like answering questions, summarizing text, and translating languages.

BERT is a foundational model built on the Transformer architecture. It led to many advancements in natural language understanding and served as a base for subsequent models. Many applications still use BERT or models heavily inspired by it for various language-related tasks.

Learn AI in 5 minutes a day.

Daily Deck explains terms like BERT as part of a free seven-card daily brief. No jargon. No fluff.

Start free