Large Language Models (LLMs) have the potential to revolutionize the insurance industry, offering new capabilities in customer service, risk assessment, and claims processing. However, they also present significant risks, such as hallucination, generating misinformation, perpetuating biases, and introducing privacy concerns. To address these challenges, GenAI solutions need to have built-in self-validation mechanism and are required to go through thorough validation process before deployment, serving as both a best practice to ensure quality of serve and a compliance measure to meet regulatory requirements.

In this presentation, Eugene, Manulife's Vice President and Global Chief Data Scientist, will explore challenges in building a reliable GenAI solution and innovative approaches in ensuring responsible AI in insurance applications, while Shone, Director of AI Validation and Governance, will share Manulife’s model risk management (MRM) practices for validating enterprise-wide generative AI solutions. Together, they will highlight the industry's commitment to developing AI systems that uphold transparency, accountability, and sustainability, using a multidimensional validation framework that addresses model risk, data privacy, cybersecurity, and ethical considerations specific to insurance operations.