AITC comments on NAIC Draft AI Principals

January 17, 2020

Commissioner Jon Godfread
Chair, Artificial Intelligence Workgroup National Association of Insurance Commissioners
444 North Capitol Street NW Suite 700 Washington, DC 20001

Attention: Denise Matthews, Director, Data Coordination and Statistical Analysis Re: NAIC Draft Principles on Artificial Intelligence Commissioner Godfread:

Thank you for the opportunity to provide public comments on the NAIC’s draft Principles on Artificial Intelligence (“draft Principles”). This comment letter is submitted on behalf of the American InsureTech Council (“AITC)”. AITC is a new organization dedicated to providing its sponsors with an independent voice on regulatory and other policy issues that are of vital interest to InsurTech’s. AITC is committed to working with the NAIC, as well as policymakers at the state and federal level to support the creation of a modern approach to regulation that is essential to this industry’s future growth.

While the draft Principles are intended to be aspirational, turning the document into actionable regulatory guidance will be a long and possibly arduous task. It should be. Our system of state regulation of insurance was developed over many decades, ensuring vital consumer protections while fostering the development of a financially stable and innovative insurance industry that is critical to the nation’s economy. Changes to existing regulatory requirements should be thoughtfully considered.> Maintaining this balance should be a top priority. The issues raised in the draft Principles are important to both consumers and insurers, and will impact the long term health of the U.S. insurance industry. We look forward to working with you throughout this process.

Listed below are a few general comments and some specific concerns.

In general terms, the draft Principles on Artificial Intelligence takes exactly the correct view. Artificial intelligence is, in and of itself, neither a pure good nor an inherent problem. It is a tool; a tool that can be employed to provide consumers with a better outcome – be it a better insurance product, enhanced policyholder convenience, lower prices, or improved claims outcomes. Some regulators have expressed concerns about the American InsureTech Council: The Future of Insurance 440 1st St., NW, Ste 430 Washington, DC 20001 use of artificial intelligence in pricing and the potential for algorithms to result in unfairly discriminatory rates or plan designs. Regulators and consumers should expect insurers to operate honestly and in good faith. We envision that as new technologies like AI change the way the insurance industry does business, some re-thinking about consumer protection standards in a digital environment will need to take place. This committee and others at the NAIC will obviously play key roles in the development of these new standards.

Fair and Ethical

We largely agree with the requirements in this section. There is no question that those utilizing AI should do so in a way that adds value to consumers. One of the hopes in using AI is that it can lead to a fairer system that roots out underlying bias, leading to more equitable outcomes and less unfair discrimination. We would , however, highlight one statement for review:

b. AI actors should proactively engage in responsible stewardship of trustworthy AI in pursuit of beneficial outcomes for consumers and society, such as augmenting human capabilities and enhancing creativity while continuing to respect cultural, social, and legal norms where they operate. The AI system should generate benefits for people that are greater than the cultural, social, and legal costs. AI systems must not be designed to harm or deceive people and should be implemented in a manner that minimizes negative outcomes. (emphasis added)

The highlighted statement is problematic as it would be nearly impossible to implement or even interpret. Which cultural, social, or legal costs should be included in the analysis? Who would act as the final arbiter of the value of those purported costs, or the value of the benefits associated with AI? We believe that a free market would exercise greater and more efficient discipline on these questions than any attempt to codify meaningful standards through regulation or legislation. We note further that this would require AI to meet a standard that the current system is incapable of meeting. We recommend deletion.

Compliant

We agree with the majority of the language in this section. It is the responsibility of the insurer to ultimately ensure not only that all rules and regulations are complied with but also that consumers are treated fairly. We do have concerns with the following language:

“Compliance with state and federal laws is an ongoing process, thus any AI system that is deployed must show consistent monitoring for compliance with laws and safeguards against outcomes that are either unfairly discriminatory or violate the agreed upon cultural, social, and legal standards”.

For the same reasons mentioned above, insurers hoping to be in compliance with the laws will find it impossible to understand how not to violate “agreed upon cultural, social, and legal standards.” Whose cultural and social standards should be control? Again, while the document is aspirational in nature, the goal should be to turn these principles into model principles and then regulatory standards. We suggest rephrasing the statement to state the following:

“Compliance with state and federal laws is an ongoing process, thus any AI system that is deployed must show consistent monitoring for compliance with laws and safeguards against outcomes that are either unfairly discriminatory or violate legal standards”.

In closing, the most important aspect of this statement of principles is that the NAIC is working towards consistency. It is vital that the states continue to work together throughout this process to ensure that the industry is not required to meet separate standards in each state. As the committee works toward a document that will operationalize these principles, it is important that the committee work diligently but without undue haste. The NAIC standard requires a super majority of states to approve a model, and that the states voting in favor of the model are committing to working toward adoption of the model in their state. We urge the states to work together to ensure a common framework.

Thank you for the opportunity to provide public comments on this important document. We look forward to working with you on these important issues.

Respectfully Submitted,

Scott Harrison (sharrison@americaninsurtech.com)
Jack Friou (jfriou@americaninsurtech.com)
The Hon. Thomas Mays (tmays@americaninsurtech.com)
Teri Hernandez (thernandez@americaninsurtech.com)
JP Wieske (jpwieske@americaninsurtech.com)

Co-Founders, The American InsureTech Council.