This is a follow up project to this paper: Automating public policy: a comparative study of conversational artificial intelligence models and human expertise in crafting briefing notes . We are finetuning a specialized AI agent for advanced policy work. The question we are answering is: Is GenAI reliable enough to tackle the “policy challenge” and does model finetuning bring us closer to the acumen of policy briefing notes crafting? A follow-up academic paper is also being drafted and has been accepted for presentation at the 7th Annual Conference of the International Association of Public Policy Association. The conference will be cohosted by Chiang Mai University School of Public Policy, from July 2 to 4, 2025, with a Pre-Conference scheduled for July 1, 2025. Note this project is deployed on Open WebUI and the resulting AI model can be found here. The results will be presented to the conference's panel " T13P04 – Advanced Computational Methods for Public Policy: NLP, ML, and LLMs chaired by: Antoine Lemor, Louis-Robert Beaulieu-Guay & Igor Tkalec. More info can be found at the 2025 ICCP Conference. Our model's generated briefing notes as well as one human-written benchmark are available in this repository.
The resulting finetuned GenAI model can be quried here.
My doctoral research is funded by Canada's Social Sciences and Humanities Research Council (SSHRC) through the Canada Vanier Graduate Scholarships