1. Gemini models simplify generative AI use cases
2. BigQuery ML allows creation, training, and execution of ML models using SQL
3. BigQuery integrates with Gemini 1.0 Pro for higher input/output scale and better quality results, expanding support for generative AI use cases
BigQuery ML allows for the creation, training, and execution of machine learning models using SQL within BigQuery. The usage of built-in ML in BigQuery saw a significant 250% year-over-year growth with customers running millions of prediction and training queries annually. Now, with the integration of Gemini 1.0 Pro via Vertex AI, users can access advanced generative AI models for tasks like text summarization and sentiment analysis directly within BigQuery using SQL statements or the DataFrame API.
With the combination of structured and unstructured data along with generative AI models, users can build data pipelines to create innovative analytical applications. For example, analyzing customer reviews in real-time and generating personalized messages and offers based on purchase history and product availability can be done within BigQuery. Plans are in place to expand support for multimodal generative AI use cases with the Gemini 1.0 Pro Vision model, allowing for image analysis and annotation using familiar SQL queries.
Unstructured data such as images, documents, and videos hold significant untapped potential within enterprises but can be challenging to interpret. BigLake helps unify data lakes and warehouses, enabling the analysis, search, security, governance, and sharing of unstructured data. New capabilities offered include the extraction of insights from documents and audio files using Vertex AI’s document processing and speech-to-text APIs for tasks like content generation, sentiment analysis, and entity extraction.
The recent integration of BigQuery vector search with Vertex AI allows for vector similarity search on BigQuery data, enabling new data and AI use cases such as semantic search, retrieval-augmented generation, text clustering, and summarization. This functionality can enhance AI models, improve context understanding, and provide more accurate recommendations, such as personalized product suggestions for online retailers. To learn more, join upcoming events like the Data Cloud Innovation Live webcast and Next ’24 for the latest developments in data and generative AI.