Google Cloud Database adds more artificial intelligence capabilities

Google Cloud is beefing up its analytic and transactional databases, including BigQuery, AlloyDB and Spanner, in an effort to drive the development of its customers' generated artificial intelligence applications.

BigQuery is Google Cloud’s advanced database service designed to support analytics and artificial intelligence tasks. The service introduces several artificial intelligence enhancements. First, Google Cloud launched a preview version of the integration of BigQuery and Vertex AI, focusing on text and speech capabilities. This integration will enable users to extract valuable insights from unstructured data such as images and documents.

Gemini, the company’s most powerful artificial intelligence model, has been made available to BigQuery customers through Vertex AI. The model sparked some controversy last week after a less-than-stellar debut on the consumer market.

These AI features follow BigQuery’s earlier announcement of vector search capabilities. The vector search feature in preview is a key component of GenAI applications, supporting similarity search and retrieval-augmented generation (RAG) using large language models.

Direct access to Vertex AI in BigQuery provides ease of use for Google Cloud AI customers in multiple ways, said Gerrit Kazmaier, general manager and vice president of data analytics at Google Cloud AI.

In the press conference, Kazmaier mentioned that as a data analysis practitioner, you can access all Vertex AI models, including Gemini models, through the SQL command line or the BigQuery embedded Python API. This makes it easier to access these models without relying on a data scientist or machine learning platform. You can access it directly in your field of work, using the data at hand. The emergence of this new technology brings more possibilities and flexibility to data analysis.

Kazmaier noted that a second advantage of integration is easier access to the data needed for AI models. Prior to integration, transferring data to AI models often required building and maintaining data pipelines to move the data. Now, he says, that's no longer necessary. "All the complicated procedures have been simplified."

In vertex, using the capabilities of text- and image-based artificial intelligence models, data analysts can gain more data analysis advantages through BigQuery to provide customers with bring more benefits.

"This opens a new stage of analyzing scenarios." He said that the summary, emotion extraction, classification, concentration, and translation of structured and unstructured data are a big deal. Roughly speaking, 90% of data is unstructured. This data is typically not used for enterprise data analytics because you can't process it in a meaningful way.

On the transactional (or operational) side, Google Cloud announced the general availability of AlloyDB AI, the AI-specific version of the hosted Postgres database the company announced at last year’s Next 23 conference. AlloyDB AI has the ability to store vector embeddings and perform vector search functions, which Google Cloud considers a core component of its customers' GenAI use cases.

Google Cloud also launched a new integration with LangChain, a popular open source framework that helps connect customer data into large language models (LLMs). Andi Gutmans, Google Cloud general manager and vice president of databases, said that all Google Cloud databases will be integrated with LangChain.

The new capabilities are in response to customer demand to find a way to get more GenAI value from their data, Gutmans said.

The company also announced that it will add vector search capabilities to other databases hosted on its cloud for customers, including Redis and MySQL. Cloud Spanner, Firestore and Bigtable will also gain vector capabilities, Gutmans said.

“What’s special about Spanner is it’s going to have nearest neighbor search capabilities, which is a slightly different variant,” Gutmans said. “What’s really exciting is the customers that have really, really big use cases. -- Trillions of vectors, for example, highly partitioned based on user. You could imagine some of Google's internal applications being partitioned by user -- they would be able to store and search vectors at trillions (vectors) scale."

"Our belief is that any database, any place that stores operational data that you might need to use in a GenAI use case, should also have vector capabilities," he said. "This is not the same as databases from 15 to 20 years ago. There is no difference when adding JSON support to both. We believe that good vector functionality should just keep the basic functionality of the database."

The above is the detailed content of Google Cloud Database adds more artificial intelligence capabilities. For more information, please follow other related articles on the PHP Chinese website!