As analytics and artificial intelligence (AI) continue to transform the insurance sector, a contingent of insurance specialists from Baker Tilly recently attended the National Association of Insurance Commissioners (NAIC) big data and AI meeting, which discussed growing trends in life insurance.
The meeting highlighted key results of NAIC’s AI/ML surveys report that examined life insurance companies across 14 states using the following criteria:
Examining the results
Of the 161 life insurance companies that responded to the survey, 58% use, plan to use, or plan to explore the use of AI/ML. Of note, this was significantly lower than the percentage of auto insurers (88%) and home insurers (70%) who answered similarly when those surveys were conducted over the past year.
The life insurance respondents who already have used or plan to explore AI/ML models are primarily using it for marketing (36%) and underwriting (34%), followed by pricing (18%) and risk management (11%). Within those four categories, AI/ML models are generally being used to augment marketing services, whereas AI/ML is typically being used for automation in underwriting, pricing and risk management.
Within marketing specifically, AI/ML most commonly is being used in connection with themes such as demographics, online media and public records. In the pricing and underwriting space, AI/ML’s most common usages are in areas such as medical data, demographics, driving behavior and credit-based insurance scores.
Of the companies that are already using AI/ML models, more than half used third parties such as LexisNexis, Milliman, TransUnion and Choreograph to develop their models.
Meanwhile, the survey respondents who do not currently use or plan to explore AI/ML cited reasons such as the absence of a business need, a lack of resources and expertise, and a reliance on legacy systems and IT.
Regulatory next steps
NAIC plans to use the data to continue assessing risk to consumers while determining whether any changes need to be made to regulatory frameworks.
Looking ahead, NAIC’s potential next steps include further exploring the decision-making connected to AI/ML models from an insurer’s perspective and also exploring the use of AI/ML at the product level. The organization also plans to evaluate the regulatory framework as it applies to the use of third-party models.
NAIC also might develop discussion topics for closed regulator-only sessions, and it will also consider whether additional AI/ML best practice documentation is needed.
Company next steps
While regulatory guidance may not be issued in the coming months, there are still steps companies should be taking to align with industry best practices and to prepare for regulation. Companies should be assessing their data governance programs, frameworks and capabilities, then aligning any potential program gaps into a data governance roadmap. Going forward, the ability to accurately illustrate the flow of data, how it is used, stored, secured and destroyed will be crucial. For those already using AI/ML, in addition to the aforementioned data governance program and framework, implementing testing criteria and key risk indicators for potential bias and/or disparate impact as well as escalation procedures for investigation will be critically important.
For more information on establishing a framework for practical data governance in the insurance industry, refer to our article on the topic.