Takeaways from the Global Insurance Symposium
Insurance industry professionals came together this week for the 2022 Global Insurance Symposium in Des Moines, which focused this year on the future of insurance and looking ahead to the next decade.
Chick Herbert, chief experience officer at NCMIC, moderated a panel Wednesday covering the emerging technology trends for insurance and what will shape them going forward.
The panelists were:
Brian Casey, partner and co-chair of Regulatory & Transactions Insurance Group, Locke Lord.
Bill Keogh, non-executive chair, RiskStream Collaborative.
Sanjeev Chaudhry, CEO, Gigaforce.
Chad Hersh, head of worldwide business and market development, life insurance industry, Amazon Web Services.
Here are some of the key points that came out of the session:
The next iteration of blockchain technologies in insurance
Keogh: What is blockchain trying to do? It’s trying to make things more efficient. I’ll give you a couple of examples of things that we’re working on. One is just simply two cars get in an accident and you have first notice of loss. Each insurance carrier, if they’re different carriers, will get a message about the claim. Typically, what happens is there are at least six phone calls between insurance carriers to make sure that each party understands what happened to the other party. It’s slow, it’s expensive, it’s not very exciting work. So our first notice of loss solution is making that happen instantaneously. That’s going to save money. It’s going to make people happier. It’s going to take out cost and low-value work.
We’re working with another company that we’re doing some integration work with that is digitizing titles for automobiles using blockchain. If you think, “Who cares about digitizing titles?” In the insurance industry we care because when you have a total loss you have to transfer the title from the insured to either the insurance company or the salvage company or both, and that takes 60 to 90 days. … They’re taking that 60 to 90 days down to as little as two hours. If you think of what happens to the asset, but it’s sitting out there in a rainy, snowy car lot waiting for the title to be transferred, the assets depreciate. So all of these things create value, and I think any kind of technology solution we’re looking at with our members is “how do we create value” or “how do we decrease costs,” and you’ve got to have a very high ROI to justify things. Everything we do is based on ROI, and I think that’s a great way of looking at things. It’s not is it a bright shiny thing, but how does it improve the business?
Applications for cryptocurrency in the insurance industry
Casey: Most people think of blockchain in terms of cryptocurrency, but from the insurance industry perspective I think there’s three applications. One is I had one client that is offshore but we are working on essentially an end-to-end vertical crypto insurance policy. What does that mean? That means there will be an insurance company that insures theft of cryptocurrency from exchanges that house cryptocurrencies. Another one would be that a life insurance company of whoever will have to take payments. I think [stablecoins] are a good bridge, and I think the younger customers in our insurance industry will demand being able to pay in cryptocurrency. And then the third area is … we’re starting to see things happen from the [insurance company’s] balance sheet side.
The future of cloud computing
Hersh: When it comes to cloud, you know, I think it’s a little bit different from some of the other things we’re talking about because it’s already so prevalent in other industries and it’s already got a really strong foothold on the [property and casualty] side of insurance. It’s hard to sometimes remember that we almost all use the cloud, maybe not every day, but at least every week. It’s something that’s become so pervasive, it’s hard to avoid using it in a big stop. But that being said, the insurance industry is really, truly at the beginning of its cloud journey. We estimate about 4% of all [the industry’s] computing workloads run on the cloud today. That’s 96% that aren’t there yet. That’s pretty remarkable.
Some of the things that are running on the cloud today for insurance are things like purpose-built apps, and there’s purpose-built databases that run much more intensively on cloud than EB2 or Oracle or SQL Server. There’s call centers and service. Amazon has something called Amazon Connect where you can set up a call center literally in 30 minutes. In 30 minutes you can have it configured and within another couple of hours, you can have it integrated into Salesforce.
If you think about 2030, I think pretty much all carriers will be building their apps in the cloud because it’s getting harder and harder to even find developers that will write apps that don’t run in the cloud. And why would you? From a cost perspective it makes little to no sense anymore. But that cost difference … that allows for innovation in a way this industry hasn’t ever seen. You don’t build something incredibly quickly, relatively cheaply, and see if it works, and if it doesn’t you just shut it down. You didn’t buy servers and so on.
The potential decentralization of internet and finance
Keogh: One interesting thing that’s emerged in our work on blockchain with the consortium is we always start by saying what we’re doing is not related to cryptocurrency. However, when you start exchanging information between counterparties, you have a unique opportunity to tokenize those things, and what that allows you to do initially is just to keep track of who’s gaining the advantage. So if I’m giving you more information than you’re giving me, we’re both benefiting but you’re benefiting more, and then how can we think about the economics of that? It doesn’t take you too far to think that eventually these could be considered transactions of value and it costs money to participate in this or you’re paying fees to be essentially saving money, so the ROI on your business case is, “OK, we’re doing this because it’s beneficial, we’re saving money, we’re charging less for the consumer, that’s a good thing.” But eventually there’s a value being created by these transactions, and I think that’s where it’s going. You don’t have to think of it as much as a currency, but just something that’s tradable.
Creating ethical artificial intelligence applications:
Casey: I think we’re just now scratching the surface of where that’s going. Regulators are doing data calls and I think the prerogative should be on them to collectively figure out if there any implicit biases [in AI applications for insurance]. My personal view is there probably are some, but I’m going to wait until I see some objectivity on that. And then how do you police it? How do you correct it? How do you monitor it? Our standard is unfair discrimination, which is basically people of the same or different class. On occasion we’d get that question, “Can we do this?” Well what’s the class? In the old days it was a question to give to the actuaries … and it’s still kind of fuzzy what everyone means. There’s not a whole lot on what class means to begin with and then it’s unfair discrimination, which basically means you can do fair discrimination. So there’s a lot in terms of nuance of whether there is discrimination at all under those standards.
Hersh: Today, biased AI is probably not a huge issue because it’s used in a very narrow way and it’s pretty easy to kind of go back and look at what the rules were preset for and determine. But as we look ahead to 2030, five to 10 years out, as the use of AI grows and it goes beyond, say, underwriting to pricing and to accurate tables and other things to get a holistic view of where that fair discrimination can be squeezed. The definition of fair may need to be looked at. You may find that it’s perfectly fair to exclude a minority group because it’s a provable causation of why their rates should be so high. In AI it can find the optimization without emotion. It doesn’t know to go ask legal or compliance, “is this OK?”
Even today, we need to be looking at baking in ethical standards into AI we build, and that’s going to be easier to do. We released a product that lets business analysts develop AI models so they can think through those things from a business perspective. But I think the one word that I hear most often when it comes to AI models is they need to be explainable. You need to go to the regulator and say this is exactly how the computer made this decision just like an underwriting manual, just like anything else we do today with humans.