Sharing how we use a cost-effective, self-hosted chatbot user interface in higher education. This session demonstrates how faculty and staff can leverage a highly customizable generative AI interface for professional development and efficiency in administrative tasks while learning about the components included in Generative AI.
Customized, fine-tuned generative AI models are revolutionizing digital learning in higher education. This session focuses on the development and implementation of an open-source chatbot interface that utilizes various Language Learning Models. By providing easy access to these models, faculty and staff are empowered to create tailored educational experiences, enhancing usability and functionality without incurring the high costs associated with multiple tools or enterprise solutions. The session will begin with a brief overview of its objectives and relevance to digital learning, followed by an introduction to the presenter and their background. The main presentation will cover the advantages of self-hosted AI models over enterprise or commercial solutions, particularly in terms of cost, customization, and data privacy. Attendees will learn about the step-by-step process of developing and deploying an open-source chatbot user interface, including integrating multiple AI models, both external and custom-created by faculty. The presentation will also demonstrate how faculty can save workspaces, presets, and prompts, and explore techniques for fine-tuning models using institution-specific knowledge bases. The creation of AI assistants to automate tasks such as lesson planning, grading, and administrative work will be highlighted, showcasing examples of how faculty can generate lesson plans, lectures, presentation slides, and supplementary materials. Additionally, staff can leverage AI for repetitive tasks, data analysis, brainstorming, and project management. Technical challenges and solutions will be discussed, addressing common issues encountered during development and deployment, along with troubleshooting strategies to ensure smooth implementation. The session will conclude with a group discussion, encouraging participants to share their experiences and challenges with implementing AI solutions at their institutions, and exploring how these strategies can be applied in their own contexts. Interactive elements will be incorporated throughout the session to engage the audience, including live demonstrations of AI models being developed and deployed, audience polls to gauge familiarity with AI and gather input on specific interests, and Q&A to provide opportunities for participants to ask questions and discuss challenges specific to their institutions. By the end of the session, attendees will understand the advantages of using in-house, self-hosted AI models for educational purposes, gain practical knowledge on developing and implementing an open-source chatbot interface, and learn how to customize AI models to fit their institution's unique needs. They will acquire strategies for automating educational and administrative tasks to improve efficiency and be equipped with troubleshooting techniques to handle technical challenges. This session offers a comprehensive guide to creating cost-effective, customized AI solutions that enhance digital learning and administrative efficiency in higher education, particularly for institutions that cannot afford expensive third-party solutions but still want to leverage the power of generative AI. Attendees will leave with actionable insights and strategies they can implement immediately to improve their own educational environments.

Revolutionizing Higher Education: Cost-Effective Digital Learning with Self-Hosted AI Chatbots
Track
Digital Learning Design and Effectiveness
Description
Track: Digital Learning Design and Effectiveness
Session Type: Education Session (45 min)
Institution Level: Higher Ed
Audience Level: All
Intended Audience: Design Thinkers, Instructional Support, Technologists, Researchers
Special Session Designation: For Instructional Designers, Presenting Original Research
Session Resource