2026 Request for Proposals: Adapting Large Language Models (LLMs) for Specialized P&C Actuarial Reasoning
Supports research adapting large language models for specialized, interpretable, and reproducible actuarial reasoning in property-casualty insurance.
Funder: Casualty Actuarial Society
Due Dates: April 27, 2026 (proposal deadline) | August 28, 2026 (interim report & executive summary) | October 2, 2026 (final deliverables)
Funding Amounts: Up to $80,000 total per project; budget includes labor and reasonable travel for presentation.
Summary: Supports research to adapt Large Language Models for specialized actuarial reasoning in property-casualty insurance, emphasizing practical, interpretable, and reproducible systems.
Key Information: All intellectual property in the final work product will be owned by the CAS.
Description
This opportunity, offered by the Casualty Actuarial Society (CAS) Artificial Intelligence Working Group, solicits research proposals focused on adapting Large Language Models (LLMs) to support core actuarial reasoning in property-casualty (P&C) insurance. Unlike prior research that uses general-purpose LLMs for actuarial tasks, this RFP seeks projects that intentionally adapt LLMs—through fine-tuning, context engineering, retrieval-augmented architectures, or hybrid models—to better align with actuarial logic, data structures, and workflows.
The goal is to develop practical, interpretable, and reproducible frameworks that practitioners can adopt, advancing actuarial practice beyond generic AI applications. Research should demonstrate clear improvements over baseline (e.g., out-of-the-box LLMs), focusing on areas such as pricing, reserving, capital modeling, reinsurance, or emerging risk assessment. Deliverables must include a research paper, code, and open-source artifacts enabling reproducibility and adoption.
