What's Happening?
At the California IT in Education Conference, Domingo Flores, IT Director of Merced County Office of Education, emphasized the benefits of hosting local large language models (LLMs) for schools. He argued that local LLMs offer greater data security and
privacy compared to external options like ChatGPT. While local LLMs may require upfront hardware costs and maintenance, they provide customization to district needs and cost efficiency. Flores acknowledged that external models might be more advanced, but the trade-off in security is worthwhile for some districts. Local LLMs can assist various departments, such as business offices with budgeting and HR with drafting job descriptions. Schools are encouraged to build proofs of concept and partner with companies like Cisco for enterprise support.
Why It's Important?
The shift towards local LLMs in educational institutions highlights the growing concern over data security and privacy. By hosting LLMs locally, schools can maintain control over sensitive information, which is crucial in protecting student and staff data. This approach also allows for tailored solutions that meet specific district needs, potentially improving operational efficiency. The move could influence other sectors to consider similar strategies for data management, emphasizing the importance of security in the digital age. Schools that adopt local LLMs may set a precedent for balancing technological advancement with privacy concerns.
What's Next?
Schools interested in local LLMs are advised to start with proofs of concept in various departments, such as business offices and HR. Partnering with companies like Cisco and NVIDIA for enterprise support is recommended to ensure successful scaling and implementation. As districts explore this option, they may face challenges related to hardware costs and expertise requirements. However, with proper support, local LLMs could become a viable solution for enhancing data security in educational settings. The ongoing development and adoption of local LLMs may lead to broader discussions on data privacy and security in other sectors.
