The many balls that the various National Association of Insurance Commissioners groups currently have in the air focusing on life insurers’ use of artificial intelligence and machine learning were reflected in the reports presented at the NAIC Summer National Meeting in Chicago:
- The Accelerated Underwriting (A) Working Group reported that it has completed work on its "Regulatory Guidance and Considerations," which was referred to the Market Conduct Examination Guidelines (D) Working Group as a basis for needed revisions to the Market Regulation Handbook.
The regulatory guidance is designed to provide a framework for regulators reviewing life insurers’ use of accelerated underwriting programs and is divided into three areas of focus: regulatory considerations, strategies for review, and request for information. The regulatory guidance is based on the NAIC’s principles on artificial intelligence and the model bulletin on the use of artificial intelligence by insurers, each adopted by the NAIC in 2020 and 2023, respectively.
In its reference to the Market Conduct Examination Working Group, the Accelerated Underwriting Working Group recognized that specific guidance pertaining to accelerated underwriting in the Market Regulation Handbook is necessary to alert market conduct examiners to the novel data and processes used by life insurers in accelerated underwriting. The referral also recommended that the handbook reference the regulatory guidance.
The Life Insurance and Annuities Committee adopted the regulatory guidance and approved the referral.
- The Life Workstream of the Special (EX) Committee on Race and Insurance reported that a survey was being developed to ask life insurers about the use of criminal history in underwriting. The workstream plans to schedule a meeting in late September to discuss the survey.
- The Third-Party Data and Models (H) Task Force was established to address regulator concerns regarding insurers’ use of third-party data and models. The task force is currently evaluating existing regulatory frameworks' utility for insurance regulators. It has been considering, for example, a “risk-focused” approach,
a “market analysis” approach, and the “Colorado” approach for developing a regulatory framework.
- The Big Data and Artificial Intelligence (H) Working Group has completed surveys of the use of AI and machine learning by personal passenger auto, homeowners, and life insurers and is now working on a health insurer survey. Also, while working on new surveys, consumer groups have called for the working group to conduct follow-up on the prior surveys. The working group reported it will first conduct target follow-up meetings with auto insurers.
The working group also heard a presentation on the limitations of the use of Bayesian Improved First Name and Surname Geocoding (BIFSG) to help identify potential racial bias in the testing of AI and machine learning models. This included an example in which BIFSG incorrectly inferred the race of the presenter.
In various meetings, Iowa Insurance Commissioner Doug Ommen spoke of the need for further collaboration among the different NAIC groups on these topics. Collaboration will certainly be essential to avoid dropping any of these many balls that the NAIC is currently juggling.
