Scouting for Growth
00:25:19 · Sabine VanderLinden
[00:25:19] S0: So much looking forward to meeting the team at the conference. And lovely to hear that you also spend a lot of time with Angela Strange. We admire Angela at ACE Extreme. And, you know, followed her and listened to a lot of talks because she's the one which who actually went into INSURTECH and observed our market as well, which I'm sure she will love this question I'm going to ask of you is we talk a lot about artificial intelligence and we know we are going to talk about AI a lot at ITC. Vegas and I'm sure a lot of the attendees, whether they are partners or insurers, chief underwriters, actuaries will want to ask you that question. How do we ensure that we use AI in a trustworthy, responsible, and ethical way when we look at all the development we are seeing in general... With generative AI?
00:26:13 · Amrit Santhirasenan
[00:26:13] S1: Yeah. It's a great question. It's a super hard question to answer. I'm really optimistic about this. Insurance is a huge judgment based game. Right? If you want to look at it through a game theoretic or or a game structured lens, there's a huge amount of judgment there that factors it. Even in the risks that are much very homogenized, very straight through algorithmically driven, judgment in selection of final rates is a really, really key thing to do. And I think one of the things that people are probably ignoring a little bit in the AI space is it's wonderful for productivity, right? It's been an amazing thing. Like most scientific and technical revolutions do, a huge driver in productivity. You know, the idea that you can ask a robot to ingest unstructured PowerPoint, photo spreadsheet. This was like complete magic even two years ago. Right? And now it's part of the course. People are frustrated if it's not 99% accurate. The idea that it's anything more than 10% accurate isn't mind blowing anymore. That's the technology for you. But I think there's two things. Think one of the other sides of AI is how it provides a framework for reasoning. The idea that you could ask a model, say, tell me why this risk you think this risk is a good thing, and it will play something play its rationale through. You could ask an AI based LLM, say, like, I'm gonna give you some information about this risk. I'm gonna give you some features about it. Can you tell me why you think this risk is good or bad? And actually, I think what we're gonna find is like, mechanisms by which we can enrich our own judgment and decision making through interaction with models is something I'm really excited about. I think you need a framework to reason about risk. You need to build a framework for operationalizing that reasoning. I'm a bit biased. I think we have the best one out there. And so we're very excited about this. With respect to the question about how we can make sure, look, I think there's a huge amount of work that needs to be done in alignment and safety and insurance specifically. I think insurance is a is a very specific kind of field of AI where actually... Sorry. AI applied to insurance is a very specific field that needs its own dedicated attention. I think insurance companies could come together. I think they should be working with technology providers.
00:28:27 · Amrit Santhirasenan, Sabine VanderLinden
We're definitely going to do that. Know, our investors are investors in OpenAI. That's cool. Right? They know this sector. I'd hope we can add something back here. We're investing very heavily in it ourselves. Virtual agents in the underwriting and actuary space are very interesting to us. We've been talking about that. We announced that at HxLive a little while ago, our first conference. We're going to be investing a lot in this space. We've got the best partners in our clients and our investors to hopefully bridge the gap and I think this is an area where I think we can contribute directly to helping build a framework for people to think about these challenges that people face around risk judgment decision making. [00:29:08] S0: I so much agree. And, you know, I often said to organization that it takes years to build, you know, AI model and generative AI models. It didn't start in November 2022. And many insurers right are working with OpenAI trying to understand what that means for their business, at least in pilot phase, right? Not in production, but everybody needs to be curious to really understand how this is going to change the world. So a fun question I hope. Let's look at future lands, Amrit. What do you think we need to start thinking about as new emerging risks, right? I am excited about space tech. As you know, there's a lot of investment going into that field. You mentioned battery and actually I heard quite a few insurers talking about battery as an asset class. So anything exciting that we need to pay attention to?
00:30:03 · Amrit Santhirasenan
[00:30:03] S1: God, what a big question. Know, all of these sectors, every single one of them an order of magnitude or two orders of magnitude more data driven than anything we've seen in the past. And in my mind, I think the most important thing... This is easy to say and really hard to do. I did some work, when Hx was much smaller with the Lloyd's lab, and we did this... We had this group called pricing for innovation. It was something we did very early when Hx was, you know, 10 people working in a car park in Southeast London, different to today. But one of the things that we talked the whole amount... Oh, thank you. Yeah. Yeah. Yeah. Little bit different from the car park, which we remember very fondly. But the new offices are great. The thing that we thought a huge amount is getting the right incentives for actually the risks being priced to the providers of the risk, the carriers of the risk being priced to share with the carriers of the risk bearing the risk. Right? So setting up that, incentive for that data to be shared between those two parties is one of the most important things to do. And I think, know, lots have talked about pricing it's a prickly space but building transparency in how that data is going to lead to better outcomes for the client I think is really important. You don't have to share every bit of your secret source to give clients and brokers much better insight as to actually how the data is being used. This is something that I think we could do much better as an industry And I think you talked a little bit about open banking, open insurance, so I know that's a controversial subject. I'm one of those people that sees the heterogeneity, the real mixed bag of data sets we see as insurance making it markedly different from much more standardized profile of what we see required to complete a banking transaction. But I do think one of the things that was done really well there was the playback of the benefits to the customer. And if you look at what happened there, how quickly now I sent some money to someone in Kyrgyzstan. It's a really random thing that I had to do, but they used the open banking protocol to do that. That wouldn't have been possible a few years ago to do that overnight to go, I need to send this money urgently to someone.
00:32:14 · Amrit Santhirasenan, Sabine VanderLinden
And I think one of the things that we could do as an industry much better when it comes to pricing frontier risks is go, actually, how are we actually gonna make that better... It better for the customer? Hard to say... Easy to say, hard to do. But, again, I'm an optimist. It's early in this field. We're investing a lot in this with some of our flagship customers and thinking about what sits around the pricing process, not just the act of going, how much does it cost? Right? And that ties intersects lots of things like explainability, in the AI space and, as well. [00:32:47] S0: And so important, right, to drive that trust through transparency with the customer we want to serve. And actually, by creating that transparency, I think we we drive retention and improve customer service. So my last question, where are we going to find you when we are going to ITC Vegas? I'm so excited. I will see you, Richard, Needy, and the team. But where can we find you? [00:33:11] S1: Gosh. I don't know the answer to that. That's embarrassing, isn't it? We're definitely gonna be in Vegas. I can find the booth out. I'll make sure when you post this that I do a link and go, sorry, Sabine, for not knowing exactly where we will be. But we will be there in force with lots of our team. I'm going to be there. I'm talking. I'm doing a talk on thinking about build versus buy decisions, particularly as they apply to AI and emerging infrastructure challenges. And so we're super excited to meet lots of people, and I'll make sure that, you know, that we share the appropriate links for this.
00:33:11 · Amrit Santhirasenan, Sabine VanderLinden
[00:33:11] S1: Gosh. I don't know the answer to that. That's embarrassing, isn't it? We're definitely gonna be in Vegas. I can find the booth out. I'll make sure when you post this that I do a link and go, sorry, Sabine, for not knowing exactly where we will be. But we will be there in force with lots of our team. I'm going to be there. I'm talking. I'm doing a talk on thinking about build versus buy decisions, particularly as they apply to AI and emerging infrastructure challenges. And so we're super excited to meet lots of people, and I'll make sure that, you know, that we share the appropriate links for this. [00:33:53] S0: I'm sure that, you know, the team is watching, and I'm sure they are going to add this in the comment box. So sorry. I didn't mean to to straddle you, but we'll add the the comment box. So people, you will know exactly where to find the team at ITC Vegas. Exactly. [00:34:10] S1: The good news is I've a very competent team. [00:34:13] S0: Indeed. Indeed. An amazing team. So thank you so much for joining me. Cannot wait to see you just in the next few days. And people, if you want to meet the Hyperexponential team, just check the comment below and reach out to the team because they are listening now. And if you want to actually set up a meeting, they will help you out as well. So thank you so much for joining me, Amrit. [00:34:38] S1: Thanks for having me, Sabine.