Automation, Artificial Intelligence Part of Plan for Modernizing Texas Insurance Department

Becoming the chief insurance regulator of a large and populous state with the second largest insurance market in the country was never in Kent Sullivan’s game plan.

But when Texas Gov. Greg Abbott asked Sullivan to step up to the plate and take over as commissioner of the Texas Department of Insurance, he willingly did so.

His predecessor, David Mattax had passed away while in office in April 2017 and Sullivan took over in September 2017 just after Hurricane Harvey struck the state to devastating effect in late August of that year.

Harvey’s aftermath obviously presented definite challenges for the new commissioner. Dealing with the “exigent circumstances” surrounding recovery from the hurricane took some time, Sullivan said. “After that, with a little bit of time to breathe and survey the situation, one of the things I found was that we really had a need to modernize the agency, to evaluate our practices in light of best practices and to upgrade. That was what we decided to do. It’s been a priority.”

The initiatives the department is working on are “pretty significant” and “will change the way we do business,” Sullivan said. However, he thinks the effort to modernize the agency’s processes will allow it to both do a better job of protecting consumers and make it easier to work with from an industry standpoint.

Review of policy forms submitted by insurance carriers is one of the agency’s major responsibilities and a process that Sullivan saw was in serious need of reform. To that extent, TDI already has begun working on a project that will enable the agency to utilize artificial intelligence (AI) to automate policy form reviews, with the aim of creating a more accurate, efficient and timely process.

In reviewing policies, “there is a commitment to try and do it on a particular time table, no surprise there. What I found was that our policy review was largely one that was a number of years old. While it’s not entirely manual, I think by modern, cutting edge, standards someone might describe it as a disproportionately manual system. It simply wasn’t the sort of system that we wanted either for speed, for efficiency of resource allocation, or for regulatory consistency purposes,” Sullivan said.

The AI system being developed is intended to monitor for regulatory consistency both vertically and horizontally, he said. “Vertical in the sense that you’re able to immediately determine the extent to which you’ve already reviewed and approved, or for that matter, disapproved, certain policy provisions that have been reviewed in the state of Texas, historically.

“It also allows you to compare that to what other states have done. That’s the horizontal approach, so that you can see what else is out there, which can be particularly useful if, for some reason, you’re visiting an issue for the first time, which comes up sometimes. You can at least see whether other states have addressed a particular issue that is raised by a policy provision.”

There are instances, as well, when a filing that has been previously addressed by regulators is re-submitted with the carrier “arguing in favor of some new interpretation. Again, it’s useful to see what other states have done. You can measure the extent to which you’re in the mainstream or perhaps not,” he said.

Texas is working on its own AI system in partnership with the National Association of Insurance Commissioners. Other states also are partnering with the NAIC on similar, but different projects. While the approaches to the problem are varied, all are aimed at streamlining and creating efficiencies in the policy form review process.

Some states are working with outside vendors. Connecticut Insurance Commissioner Katherine L. Wade, for example, announced in late December 2018 that her department is working on a pilot with RiskGenius, a Kansas-based tech company. The purpose of that project is “to test a web-based platform that applies machine learning to determine if it improves the regulatory reviews of insurance company filings and ultimately moves approved products more quickly to the market for greater consumer choice,” according to the Connecticut Insurance Department.

In a sense, the states are playing catch up with the industry, which already is using sophisticated AI tools in a variety of ways, Sullivan said. Ultimately, as the various processes being tested are finely tuned, regulators will have multiple alternative AI form review options from which to choose. As for the Texas project, he said TDI anticipates “a proof of concept here just in the next couple of months. Then we hope to be able to implement it within a reasonable time frame and it’s something that probably will be ultimately available across the country.”

In addition to the benefits of efficiency and accuracy, Sullivan sees a consumer protection aspect to using artificial intelligence in the policy form review process.

“You make better regulatory decisions when you have access to the maximum amount of information. That’s what we’re really trying to do with the AI project, is to ensure that you’re delivering consistent decisions about issues raised by policy forms and also have access to what other states are doing,” he said.