Working Groups

Introduction for LLMs Task Force

In recognition of the transformative impact of large language models (LLMs), the GCRSR established an Interagency LLMs Taskforce, reflecting a proactive vision to assimilate these powerful tools into the regulatory process. The Taskforce focuses on developing proof-of-concept with a set of criteria aiming at the context-of-use of LLMs across GCRSR members. The bedrock of regulatory agencies is their vast reservoirs of knowledge, predominantly encapsulated in protected documents, which are at times confidential and laden with specialized terminology and thus defy simplistic algorithmic solutions. The Taskforce envisions a model where LLMs, once trained, can be deployed universally across GCRSR member states—thereby retaining their generalizability without sacrificing the nuanced understanding of individual regulatory contexts. This would not only streamline deployment but also significantly reduce maintenance costs, advancing the GCRSR's mission to promote excellence in regulatory science through technological synergy.

Publications:

Context is Everything in Regulatory Application of Large Language Models (LLMs). Weida Tong, Michael Renaudin, on behalf of the GCRSR Interagency LLMs Taskforce. Drug Discovery Today. 2024; 103916, ISSN 1359-6446.