Skip to main content Skip to navigation

Disrupt Your Research Using Generative AI Powered ScienceSage

Project Overview

The document examines the role of generative AI in education through the implementation of ScienceSage, a web application that utilizes large language models (LLMs) to boost research productivity across scientific disciplines. ScienceSage enables users to create and manage knowledge bases with multimodal data, facilitating the generation of detailed research reports and allowing for interactive inquiries related to uploaded documents. It emphasizes the potential of this technology to enhance the efficiency and breadth of research and product innovation within educational organizations. By streamlining the research process, ScienceSage aims to significantly elevate the quality and speed of academic inquiry, ultimately transforming how educators and researchers interact with data and develop new ideas. The findings suggest that generative AI can play a pivotal role in modernizing educational practices, making them more responsive to the needs of contemporary research environments.

Key Applications

ScienceSage

Context: Used by researchers in various scientific fields within a consumer packaged goods company to extract insights from multimodal data.

Implementation: Deployed as a web application on a GPU server, utilizing various indexing methods (vector index, knowledge graph index, custom index) to support data retrieval and report generation.

Outcomes: Enhanced speed and efficiency in research projects, enabling quick iteration of ideas, search, and summarization of information.

Challenges: Managing compatibility across different embedding databases and LLM versions.

Implementation Barriers

Technical Barrier

Challenges in managing multiple embedding-based databases and ensuring compatibility across different versions of LLM packages.

Proposed Solutions: Streamline to a single database system that can handle all indices; improve backward compatibility among evolving platforms.

Project Team

Yong Zhang

Researcher

Eric Herrison Gyamfi

Researcher

Kelly Anderson

Researcher

Sasha Roberts

Researcher

Matt Barker

Researcher

Contact Information

For information about the paper, please contact the authors.

Authors: Yong Zhang, Eric Herrison Gyamfi, Kelly Anderson, Sasha Roberts, Matt Barker

Source Publication: View Original PaperLink opens in a new window

Project Contact: Dr. Jianhua Yang

LLM Model Version: gpt-4o-mini-2024-07-18

Analysis Provider: Openai

Let us know you agree to cookies