Tokio Marine HCC, the Houston-based U.S. specialty insurance arm of the Japanese insurer, is using a microservices approach to implementing AI, according to Arron Lamp, CIO of the Public Risk Group at the U.S. unit.
Lamp, speaking at the recent
"When you figure out what you want done, you put it into something deterministic that lets it actually process through the system," Lamp said.
Insurers often mention
"I need to be able to define what went in and why it went in, what came out when it came out, and was it accurate," he said. "If you can't build that idea that I can run it and not have to go back and check it myself, you're not adding much value."
Controlling the data is necessary to ensure operations staff are accessing the correct data for certain types of transactions, according to Lamp. "You can't have AI just randomly picking what index the right data is," he said.
The narrow focus of a microservices AI implementation makes human oversight or review easier, Lamp explained.
"I can't use it if I can't put something in front of my compliance officer, that there is a human in the loop that's actually going to see the results and say it doesn't make sense," he said. "That's a big piece of how we're thinking about AI coming into our environment and actually being a very big help to the organization at production scale."
Insurers need to reduce risk in their own use of AI, as Lamp's colleague, Robert Pick, EVP and CIO of Tokio Marine North America Services, the insurer's professional services arm,
"Too often I read that AI is something like a solar flare or asteroid, something happening to us," Pick said. "AI is something that we are doing to ourselves. We control the pace, we develop the pace as a society, as a technology industry, as an insurance industry. Once you reverse that flow, it's pretty easy to get a sense of control back."
