BigQuery’s generative AI functions, AI.GENERATE and AI.GENERATE_TABLE, are now generally available (GA), enhancing data analysis capabilities. These functions accept diverse input types, including text, images, audio, and documents, allowing seamless integration into SQL queries like SELECT, WHERE, and ORDER BY clauses for advanced data processing. AI.GENERATE excels in free-form text generation, making it ideal for tasks such as summarization, translation, and sentiment analysis through simple prompts.
It also supports structured output, enabling users to define output schema for clearer insights. By specifying output field names—like “sentiment” or “summarize_in_one_sentence”—multiple AI tasks can be executed concurrently with a single function call, making results easy to use in subsequent applications. For instance, an example showcases AI.GENERATE performing five tasks—entity extraction, topic modeling, sentiment analysis, translation, and summarization—simultaneously on text data, demonstrating its robust capabilities in data analytics.
Source link
