Retrieval-Augmented Generation (RAG) #1
Description
Description
Adds Retrieval-Augmented Generation (RAG), with potential to help students solve more complex problems. This entails giving Firebot the ability to retrieve information from documents, such as course materials (with instructor permission), research papers, syllabi, etc.
Targeted information provided by Firebot can then encourage deeper inquiry. Furthermore, use of specific, pre-vetted, and reputable resources can demystify the "black box" nature of many AI systems, thus providing the student-user better insights. The most valuable improvement, though, would be that Firebot would be able to hold the student to Pacific Ridge School's high academic standards.
Implementation steps
- Implement a retrieval mechanism: This could involve searching through existing documents or an external knowledge base based on the user's query.
- Augment the prompt: Incorporate the retrieved information into the prompt sent to the language model.
- Modify the system prompt: Update the system prompt (.../prs-firebot/lib/ai/prompts.ts) to instruct the model on how to use the retrieved information.
- Potentially create a new tool: A new tool could be created to handle the retrieval and prompt augmentation steps.