Google Deploys BERT Neural Network for Major Search Algorithm Update
Google rolls out a significant search update using BERT, a new neural network technique that impacts one in ten English queries in the U.S. to better understand conversational search intent.
Google announces one of its largest search algorithm updates in recent years by integrating a neural network technique called BERT. This new system affects approximately one in ten searches in the United States for English queries and is already live globally for featured snippets. Such a wide-reaching change is highly significant in the search industry and poses new challenges for search engine optimization experts.
The technology, known as Bidirectional Encoder Representations from Transformers, excels at understanding the sequence of words in natural language. Because of this, the update drastically improves Google's ability to handle longer, conversational queries where interpreting full sentences matters more than matching isolated keywords. Google first introduced and open-sourced this machine learning model last year.
This rollout also marks the first time Google uses its latest Tensor Processing Unit (TPU) chips to serve search results at scale. The combination of advanced hardware and smarter language processing allows the search engine to grasp the exact intent behind user queries. The update launches this week, meaning users currently experience more relevant search results and featured snippets.