Understanding Instructor: Elixir's Gateway to AI Integration

Thomas Millar presented Instructor_ex, a project that seeks to make it easier for Elixir developers to integrate artificial intelligence (AI) capabilities into their applications. One of the key challenges in incorporating AI, particularly machine learning models, into existing Elixir systems is the difficulty in parsing and validating the unstructured string data that these models often return. Instructor_ex tackles this issue by utilizing Ecto schemas — a feature of the Elixir ecosystem commonly used for defining and validating data structures — to define the expected format of the AI model's output, thereby ensuring consistency and simplifying integration. Instructor_ex is not limited to open AI function calling, as it also works with other AI models like Llama and Bumblebee. Millar delves into use cases such as text classification, entity extraction, data generation, and even automating project management tasks like extracting action items from meeting transcripts.