MassMutual

MassMutual deployed an algorithmic underwriting system called the MassMutual Mortality Score (M3S) that uses machine learning to predict the mortality risk of life insurance applicants, replacing a points-based manual review process for most policies.

The M3S system uses survival modeling trained on millions of historical life insurance applications spanning 15 years, incorporating health, behavioral, and financial data to generate a mortality risk score for each applicant. According to a peer-reviewed paper published at AAAI by MassMutual data scientists, the system reduced time to issue policies by more than 25% and increased customer acceptance rates by more than 30% for offers made with light manual review. The underlying mortality model also underpinned LifeScore360, a product offered to other insurers via a now-discontinued subsidiary (LifeScore Labs); however, M3S continues to underpin MassMutual's own internal underwriting process.

MassMutual integrated Microsoft's Copilot AI assistant and GitHub Copilot code-completion tool into its employee productivity and software development workflows, making them available to employees after mandatory AI training.

According to MassMutual CIO Sears Merritt, the company integrated generative AI capabilities broadly into all productivity and collaboration tools, including Microsoft 365 and Zoom. Specifically confirmed third-party AI tools include Microsoft Copilot (for general productivity) and GitHub Copilot (for software code completion). Employees must complete formal training before gaining access, which covers what generative AI can and cannot do and its risks. These tools are used internally by employees only; policyholders do not have access.

MassMutual deployed internal AI virtual assistants that help customer service representatives and financial professionals answer questions about insurance policies, including older policies created decades ago.

According to MassMutual's CIO Sears Merritt, the company built and deployed these AI virtual assistants in approximately nine months. The assistants use a ChatGPT-like conversational interface to help employees retrieve answers to policy-related questions. As of February 2025, the tool had reduced customer service response times by 10% and cut by 5% the number of calls financial advisors place into a contact center. Human employees remain in the loop for all decisions. As of February 2025, all generative AI products at MassMutual were restricted to internal employee use, with no plans to expose them directly to policyholders.