Generative AI in P&C: Transforming Claims and Customer Service

Abstract

Generative Artificial Intelligence (GenAI) is rapidly becoming a breakthrough in the sphere of property and casualty (P&C) insurance on two levels: claims and customer support.  The current paper discusses how the idea of generative models and other AI-related approaches (large language models, multimodal generation, retrieval-augmented generation, etc.) can complement, automate, and redesign the steps in the claims lifecycle and customer interaction process. We introduce a GenAI infrastructure and approach into main P&C operations, and run emulated deployments of GenAI in claim acquisition, document processing, fraud detection, settlement bargaining, and chat-based customer support. We found that GenAI can save on turnaround time (reduction by approximately 3050) and loss adjustment costs (reduction by approximately 20) and will notice a significant increase in customer satisfaction rates. Risks, regulatory constraints, interpretability of the model and challenges of deployment are also discussed. Lastly, we discuss areas to explore in the future, such as federated learning among insurers, generation that is contract aware, multimodal damage evaluation, and trust architectures. The evidence established that generative AI can revolutionize P&C claims and service on a large scale, though it is important to consider human controls, ethical boundary, and domain adaptation