Cloudera Docs

Advantages and use cases of Impala AI functions

Learn how you can use Impala's ai_generate_text function to access Large Language Models (LLMs) in SQL queries. This function enables you to input a prompt, retrieve the LLM response, and include it in results. You can create custom UDFs for complex tasks like sentiment analysis and translation.

Use LLMs directly in SQL with Impala's ai_generate_text function

Impala introduces a built-in AI function called ai_generate_text that enables direct access to and utilization of Large Language Models (LLMs) in SQL queries. With this function, you can input a prompt, which may include data. The function communicates with a supported LLM endpoint, sends the prompt, retrieves the response, and includes it in the query result.

Alternatively, seamlessly integrate LLM intelligence into your Impala workflow by creating custom User Defined Functions (UDFs) on top of ai_generate_text. This allows you to use concise SQL statements for sending prompts to an LLM and receiving responses. You can define UDFs for complex tasks like sentiment analysis, language translation, and generative contextual analysis.

For examples of AI built-in function arguments and how to create and use custom UDFs with the built-in AI function, see Overview of AI function syntax and arguments.

Advantages of using AI functions

  • Simplified Workflow: Eliminates the necessity for setting up intricate data pipelines.
  • No ML Expertise Required: No specialized machine learning skills are needed.
  • Swift Decision-Making: Enables faster insights on the data, facilitating critical business decisions by using in-database function calls.
  • Integrated Functionality: Requires no external applications, as it is a built-in feature in Data Warehouse.

List of possible use cases

Here are some practical applications of using AI models with the function:
  • Sentiment Analysis: Use the AI model to examine customer reviews for a product and identify their sentiment as positive, negative, or neutral.
  • Language Translation: Translate product reviews written in different languages to understand customer feedback from various regions.
  • Generative Contextual Analysis: Generate detailed reports and insights on various topics based on provided data.