We are excited to announce a groundbreaking service that leverages the power of generative AI to transform how LSA departments manage and disseminate information. The LSA Department GenAI Assistant project, powered by the U-M Maizey GPT platform developed by U-M Information and Technology Services (ITS), aims to equip departments with a toolkit to create AI-driven online assistants tailored to their specific needs. By implementing this technology, LSA units can streamline responses to inquiries, enhance user engagement, and reduce the workload for department staff.
Project Overview
U-M Maizey GPT, developed and supported by ITS, is a generative AI platform available to all active U-M faculty, staff, and students on the Ann Arbor, Flint, and Dearborn campuses and Michigan Medicine. This tool allows U-M faculty, staff, and students to enrich their GenAI experience based on a custom dataset that they provide. This service empowers users to extract valuable insights, discover patterns, and gain deeper knowledge from the available datasets.
The LSA Department GenAI Assistant project was conceived to help departments implement this technology across multiple datasets. This will address a common challenge faced by many departments: the substantial time and effort spent responding to inquiries from prospective students, family members, donors, alumni, and other stakeholders. Often, the information sought is readily available on department websites, yet the task of directing individuals to the right resources remains labor-intensive. By leveraging Maizey to develop a department-specific AI assistant, LSA departments and units can provide instant, accurate responses to users, thereby freeing up staff to focus on more complex and value-added tasks.
The project began in June of 2024 with an ambitious timeline that culminated in the launch of the toolkit last month. Key milestones included holding a kickoff meeting, developing a prototype AI assistant, crafting customer service and support best practices, and creating and disseminating the toolkit to all LSA departments.
Benefits of the GenAI Assistant
The primary advantage of the GenAI Assistant is its ability to provide instant, 24/7 responses to inquiries based on publicly available department web content. This ensures that users have access to accurate information at any time, without the need to wait for office hours or staff availability. Departments will benefit from reduced email and phone inquiries, allowing staff to focus on strategic initiatives and personalized interactions that require human touch.
Toolkit and Implementation Process
The GenAI Assistant Toolkit is a comprehensive guide designed to help departments create and integrate their own AI assistants using the U-M Maizey GPT platform. It includes detailed instructions on configuring the AI, connecting data sources, and best practices for ensuring accurate and helpful responses. Key steps in the process include:
Creating MCommunity groups to manage administrative and user access.
Configuring AI settings by adjusting various parameters like “prompts,” “chunks” and “temperatures” to fine-tune the AI's performance.
Integrating Google Drive folders, public website URLs, and other data sources that AI will use to generate responses.
Conducting extensive testing to ensure AI provides accurate and relevant answers.
Launching the AI Assistant by making it publicly available and promoting it to the intended audience.
Use Case Examples
To illustrate the practical applications of the GenAI Assistant, two prototypes were developed for the Departments of Psychology and Ecology & Evolutionary Biology (EEB) departments. The Psychology department's assistant, named Sigmund, directs undergraduate students to relevant web pages, while EEB’s assistant, REEBA, is behind a Weblogin and helps graduate students navigate departmental policies and procedures. These prototypes serve as models for other departments to customize their AI assistants according to their unique needs.
Recognizing Contributions
We extend our heartfelt appreciation to all the individuals who have contributed to the success of this project. Special thanks to our project sponsors, subject matter experts, technical leads, and the ITS GenAI team for their unwavering support and collaboration.
This project would not have been possible without the vision and support of project sponsors Cathy Curley and Merlis Nolan. Additionally, the expertise and dedication of our department subject matter experts—Chris Gale (Physics), Karly Mitchell (Economics), Amy Rundquist (Anthropology), Kaydee Szust (Psychology), and Rory Walsh (EEB)—have been instrumental in shaping the project. Our technical lead, Jeremiah Cook, and the ITS GenAI platform contact, Ben Andries, have provided invaluable technical guidance and support throughout the project.
Get Started with Your GenAI Assistant
We invite all LSA departments to take advantage of this innovative service and explore the potential of AI to enhance user engagement and operational efficiency. The toolkit is now available, and our team is ready to assist you in setting up your department's AI assistant. For more information or assistance, please complete the following form.
Join us in revolutionizing how we connect with our stakeholders and provide timely, accurate information. Together, we can harness the power of AI to create a more efficient and responsive university environment!
