About the Initiative
A comprehensive project to build an AI-powered library assistant that transforms academic research at St. Cloud State University.
Our Mission
The HuskyBot — SCSU Library AI aims to develop a state-of-the-art, custom-trained large language model (LLM) specifically designed for academic research support. This project represents a bold step toward integrating artificial intelligence into the university library ecosystem.
Our hybrid approach combines a user-friendly chatbot interface with the deep knowledge capabilities of a fine-tuned LLM, trained on curated academic datasets from ArXiv, Semantic Scholar, and PubMed. The result is a powerful research companion that understands the nuances of academic writing and scholarly research.
Beyond simple question-answering, the assistant will serve as a comprehensive research paper writing tool — helping with literature reviews, citation generation, thesis development, methodology suggestions, and draft review. It’s designed to enhance the research process while maintaining the highest standards of academic integrity.
Project Goals
Six core objectives driving the development of the HuskyBot.
Democratize Research
Make powerful AI research tools accessible to all students and faculty, regardless of technical background.
Hybrid Interface
Combine the conversational simplicity of a chatbot with the deep analytical power of a fine-tuned LLM.
Writing Excellence
Provide end-to-end research paper writing support from topic exploration to final draft review.
Community Building
Create a knowledge-sharing platform that connects researchers across departments and disciplines.
Innovation Hub
Position SCSU at the forefront of AI-assisted academic research in the Minnesota State system.
Academic Integrity
Design the assistant to promote learning and understanding, not replace critical thinking or original work.
Deployment Model
Web-Accessible
Browser-based interface accessible from any device on campus or remotely, integrated with SCSU’s existing library systems.
Downloadable Model
Open-weight model available for local deployment, enabling offline research and customization by departments.