Oracle has announced plans to add the vector search feature to its Oracle Database 23c software, which is designed to support the increased usage of large language models (LLMs). Typically, AI services come with embedding services that convert text into vectors, enabling the development of applications that can search for similar content by calculating the distances between different vectors. This shift in database thinking is due to the inability to directly index a dataset. The need to query data in vector form has led to the creation of new databases that can handle vector-like data, such as PostgreSQL’s pgvector. While Oracle Database 23c has been in the GA stage since September 19th, the full implementation of the vector feature will be added at a later date.
TLDR: Oracle plans to enhance its Oracle Database software with a vector search feature to accommodate the rising use of large language models. This change in database thinking is due to the need to query data in vector form, as traditional indexing methods are insufficient. While Oracle Database 23c is currently available, the vector feature will be added in the future.