Takeaways:
- Plug-ins are catching on for insurers' actuarial operation needs
- AI can catch problems with customer service interactions live, in the moment
- Natural language processing and transfer learning can highlight data issues
Amanda Turcotte, senior vice president and chief actuary at Amalgamated Life Insurance Company, joined the insurer in 2023 after executive roles including pricing and underwriting functions at Guardian, Prudential, Chubb, AXA and Equitable. She has seen the concept of insurtech gradually catch on over the past 10 years in the industry, and has firsthand insurtech experience as chief product officer of Ostraa (now Resilient), a mobile platform for insurance distribution, from 2016 to 2019. Digital Insurance spoke with Turcotte about transformations in insurance, particularly in actuarial operations.
Editor's Note: This article is from a longer interview and edited for clarity.
How has insurtech affected actuary work for insurers?
Insurtech has changed dramatically since 2016. At the beginning it was focused on disruption, digital transformation, especially on the property and casualty side. A lot of insurtechs were looking to displace traditional insurers.
Today, we've really changed the focus of insurtech to B2B, where insurtechs are very happy to collaborate with traditional insurance carriers. They've learned that being an insurance carrier is not the easiest thing in the world to do.
There's a lot of details to running a full stack insurance company. It's not the vehicle that a lot of insurtech entrepreneurs are looking for anymore. It's really about building ecosystems within insurance carriers, and partnering to find those niches within a carrier where you can add value.
We've seen that evolution on the actuarial side as well. There were some early disruptors on modeling platforms. For example, Slope is great modeling software that was an insurtech startup coming in and disrupting a lot of the existing modeling platforms. If you go to any actuarial convention, there's a lot of opportunities to have plug-ins that give you a technology boost. It really lowers the barrier of entry for actuaries to access these insurtech tools.
Replacing a modeling platform or implementing a huge new system is a huge corporate decision that's going to take three to four years from first conversation to final implementation. Now those conversations can be made at a lower level in the company, because there are six month projects and they add incremental value to the activities we do.
Have actuarial plug-ins replaced using one large insurtech for all of an insurer's needs?
I don't think so necessarily. A lot of carriers weren't engaging with insurtechs when it was just a huge, big platform shift. Often you had a small startup approaching a large, established company. We had this experience at Equitable, approaching a carrier with all sorts of security protocols and various legacy system integrations. Some have APIs, some do not.
Those were barriers that created a lot of friction for some of the early insurtechs to really collaborate. That's why we've seen the big shift towards incremental, widgetized, plug and play insurtechs. That's not just exclusive to actuarial work. We see this in underwriting and claims and customer service, where there's not an overhaul of your whole system, but rather using an agent that can plug into a process and save hours per case.
Is regulation addressing the use of AI to power such plug-ins or agents?
It is the focus of a lot of development these days. Insurance is highly regulated. This always adds complexity to the process. Regulators felt a little bit behind the curve. With technology emerging in the mid 2000s, even into the 2010s, there were regulators who were very on top of emerging AI development, even ahead of where the industry was, in states like Connecticut, Colorado, California and New York. Now a lot of states have AI regulation, which tells you what you can and can't do.
A lot of that regulation was drafted in the very early days of AI when the industry hadn't quite figured out how it wanted to interact with AI tools. The regulations are challenged to keep up with the speed of innovation and the sophistication of some of the AI tools in the market. If regulation says you have to demonstrate 20 items for any AI that you put into your underwriting, customer service or claims processing, and corporate sponsors want to implement those functions, someone from your regulatory and compliance team will ask if you have a solution for those 20 items. It does end up creating some friction. The big carriers have started with very select deployments of in-production AI solutions.
