This project is actively unfolding and will continue through 2024. During the Open Education Day, we'll present the insights and discoveries we've gathered up to that point.

This presentation will offer a dive into the practicalities, challenges, and decisions involved in implementing large language models (LLMs) within higher education institutions. We also aim to provide attendees with a clear understanding of the potential and pitfalls of local LLMs compared to cloud-based solutions.

1. Why Do We Need Local Large Language Models (LLMs)?
- Privacy: Emphasizing the importance of data confidentiality and secure handling of sensitive information in the educational sector.
- Bias: Exploring strategies to recognize and mitigate inherent biases in LLMs for fair and unbiased educational outcomes.
- Ethical Concerns: Discussing the ethical implications of AI in education to ensure responsible and equitable technology use.

2. Requirements for Implementing Local LLMs
- Hardware: Outlining the computational and storage infrastructure necessary for hosting LLMs locally.
- Software: Reviewing software requirements, including LLM platforms and tools for effective deployment.
- Integration in Existing Systems: Strategies for integrating LLMs with existing educational systems like Moodle.

3. Comparative Analysis with Other Solutions
- Software as a Service (SaaS): Comparing the benefits and limitations of cloud-based LLMs versus local setups.
- Infrastructure as a Service (IaaS): Assessing how local LLMs compare against infrastructure-focused solutions in cost, scalability, and customization.
- Providing a comprehensive overview to assist educators and administrators in making informed LLM technology adoption decisions.