Title: Harnessing Large Language Models (LLMs) for Traceable Results

Abstract: When working with public sector clients we must balance the power of cutting-edge machine learning tools with our clients’ need for transparency and compliance. This balance is particularly difficult to strike when working with the industry’s latest AI darling: large language models (LLMs). The media frenzy about generative AI (of which LLMs are the motor) means they are both alluring and alarming as a proposed solution, and clients that value transparency and traceability will have plenty of qualms about their application.

To this end, we will provide clarification on LLMs, generative AI, and their respective applications to ensure a baseline of understanding for all audience members. We will then elaborate on our balanced technical solution to inform other practitioners working in similar contexts and other public sector entities with similar business needs. In our fully open-source models, we integrate LLMs within more controlled rules-based models, to leverage the power of the latest technology while ensuring a high degree of traceability for the client and its auditors.

Watch the full video of our talk at the GovAI Summit here: