AI in the Insurance Industry: The good, the bad and the unknown

Enjoy complimentary access to top ideas and insights — selected by our editors.

Patti Harman (00:06):

The introduction of artificial intelligence into the insurance industry has been met with a mix of both concern and excitement. The ability to streamline manual, time-consuming tasks, sort through vast amounts of data to improve underwriting accuracy and binding policies faster as well as improve customer service at every stage of the insurance process is extremely appealing to carriers, agents, and brokers. At the same time, providing access to personally identifiable information and other confidential data creates some risks for insurers and vendors who handle these documents, particularly as compliance regulations evolve around the globe. It would probably surprise a lot of people to learn that we’ve all been using some form of artificial intelligence for well over two decades, but we’ve probably become more aware of its use as it’s become more mainstream, powering chatbots, analyzing weather patterns to predict future forecasts or helping companies to better manage their marketing and customer service activities. Joining me today to discuss all of these issues and much more is Garrett Gray, president of Global Insurance Solutions at CoreLogic, where he leads a team of industry experts focused on building market-driven solutions across the property insurance ecosystem. Previously, he was the founder and the CEO of NextGear Solutions, which was acquired by CoreLogic in August, 2021. Thank you so much for joining us today, Garrett.

Garret Gray (01:37):

Well, thanks for having me. I’m excited to be here.

Patti Harman (01:39):

So in my opening, I mentioned a few uses for AI, like the chat bots and weather predictions. I probably should have added in like GPS for directions, that sort of thing. But I would like us to go a little bit deeper and chat about what AI is and some of the benefits that it provides to the businesses and the industries that are adopting it.

Garret Gray (02:03):

And you made a really good point in your opening that AI has really been around for quite a long time. So especially at CoreLogic, we have been at the forefront of using AI in all of our products, whether clients knew it or not. It’s one of those things that we’re just starting to talk about now because now people are starting to understand it. It’s hit the public consciousness in the last 12, 13 months and with all the different versions of chat GPT and all the image creators and the video deepfake creators, it is just become a huge topic everywhere. And so it now makes a lot more sense for us to kind of explain a little bit of the magic behind a lot of the projects that we have because they’ve been AI-driven for a long time. So AI is essentially a tool that we can use to make computers smarter and to make machines smarter. And essentially, the basic premise is things that require a lot of calculation and are mundane that humans have to do, can really be automated in such a way that they are given insights to make better decisions faster. And so there is a lot of ways that AI can be implemented and people have heard about Generative AI and how it can create these documents or these articles out of whole cloth. Sometimes they’re great, sometimes they’re not. But in insurance, it’s really starting to impact every part of the ecosystem.

Patti Harman (03:38):

Very true. It’s funny, as an editor, I get a lot of pitches from different people and somebody sent me one and I’m like, this is not how people write. This isn’t how they speak. And it hit me. This was generated with AI, and so my response back was, next time write it yourself and don’t have your AI send it to me. And I never heard from them again. So there are some very specific applications for the property and casualty industry as well, and I know you cover some of those in the new report that CoreLogic has coming out, the role of artificial intelligence across the property insurance ecosphere, would you share some of the ways carriers and other insurance-related businesses are using AI?

Garret Gray (04:21):

Yeah, so there’s lots of different use cases, but I’ll give you a couple of the ones I’m most excited about right now. There is a way that we’ve developed a few years ago that is now becoming a little more popular of determining the first floor height of a property. So if you wanted to understand what the risk was for flood, you wanted to map that against the local floodplains. Understanding the actual first floor height of that property is really important, and getting those certifications have been really difficult for carriers. So we can automate that because we have all the exterior imagery to analyze and then use AI to determine how high the first floor height is. So it’s like things like being able to mine photos and imagery and extract data that can be used to help carriers assess risk faster and more accurately and help insurers have more access to insurance because carriers can make more targeted decisions is one of the really exciting use cases on the underwriting side for AI.

(05:29):

On the claim side, there’s a lot of work that goes into adjusters trying to sift through thousands of line items on an estimate and make sure the scope and what the contractor and the supply chain is doing is appropriate to get a property back to pre-loss condition. That whole analytics can be automated through AI today, and so adjusters can be served up, Hey, here are the things you should look at. Here are the things that seem completely perfect and get to faster fair settlements for everybody to speed up the process to get people back in their homes faster. And so those are just two examples, but if you can imagine the entire workflow of both pricing, insurance and then responding to a claim when there is one, every part of that, if there’s a manual kind of laborious step, AI is really starting to touch every piece of that.

Patti Harman (06:28):

That’s really exciting, especially in the claims space because you’re right, there are just so many details that they’re trying to track and for it to sift through all that information and flag what they should be watching or at least double-checking that sort of thing, no one’s going to really complain that, oh, this is going to save me a whole lot of time. I don’t think when we talk about using AI and automation in terms of insurance, I think it’s easy to kind of confuse their roles. Could you share with our audience the differences between how automation and AI are used?

Garret Gray (07:03):

Yeah, I mean, I think there is some connection. So AI is the kind of main intent is to automate something. And there have been different ways of automating different tasks through different algorithms. And I think what’s different about AI is it’s not just automating mundane tasks, although it is doing that, it’s actually providing intelligence and it’s taking a ton of data, and that’s actually a really important point. The quality of the data that the AI is trained on, it really influences the outcomes. And so when you’re thinking about who is sitting on the right type of data to have really unique quality and perspective to then train AI to get to better outcomes, this is something that is different than just automating a task. This is deep insights, predictions, making sure that things that humans would typically have to think through and actually like logic through can be done in a faster way.

Patti Harman (08:15):

It would make things a lot easier for people. And that was a really good explanation of the difference between the two and how they’re using it. What are some of the most important factors for organizations to consider when they’re implementing an AI-focused business strategy? You talked about the importance of good data and that sort of thing, and then there’s some pitfalls that they should be looking at as well.

Garret Gray (08:38):

So a couple of things to consider. One, the way that AI is trained is on data. So you want to pick a partner that you think has really good data, and I think that’s really important. The other thing is we have a particular philosophy at CoreLogic on how we’re implementing AI. There’s a lot of companies who are doing some exciting things, but they’re a little bit ahead of the technology, so maybe they’re going to do something transformational in the next five years, but it’s hard to implement something like that that’s so transformational for carriers. And so our approach has been, let’s take the workflows that have already existed and let’s incrementally use AI to make them smarter and faster so that when you wake up five years from now, it is transformational, but you were getting the value along the way as opposed to waiting for this really cool tech to be available and actually work in most cases.

(09:37):

And we’re still, so certain things like image to scope, which a lot of us are working on, we think it’s going to be here in a few years and we’re doing a lot of that today, but some are focused on saying, Hey, we need to get fully there before we can implement it. We think, Hey, we can actually take images and start automating pieces of the scope even before we’re fully to image to scope. And that gives people value every day and also lets them kind of ease into the technology, which makes it easier to implement.

Patti Harman (10:09):

Wow. Are there certain things that they should keep in mind in terms of as they’re going through the implementation strategy? Are there certain pitfalls that you want them to keep in mind?

Garret Gray (10:20):

And I know I keep harping on data, but if you aren’t really comfortable with the outcomes you’re going to get, if there’s biases that are built in, you got to really watch those sorts of things. Our company has really put in some strict governance around that so that we can make sure that whenever those need to be tested, that we have the right documentation and we can show how we’re making decisions. And I think that’s another important thing. Carriers need to understand who they’re partnering with, how they’re training their models, what data underlines it, and if there’s any biases that are potentially being introduced and make sure that is taken care of and thought through ahead of time.

Patti Harman (11:06):

Yeah, I’ve spoken to a number of carriers and that is definitely something that’s top of mind for them is making sure that the information that they have, both the input and the output is just really accurate. We’ve seen the rise of insurtech over the last decade or more, and I think the pandemic probably really helped to spur its adoption as companies realize they had to adjust their operations as everyone transitioned to the remote work environments. What are some of the AI-based solutions the insurance industry has adopted or at least begun to embrace over the last several years, and how important is it for them to be proactive in using these technologies?

Garret Gray (11:47):

Yeah, I think Covid and the Pandemic is a really good place to start because the insurance community was moving down this kind of train track already. They were moving towards more automation, more AI empowered workflows that digitized the process and allowed maybe less adjusting out in the field and more adjusting at the desk. And so that was happening, but Covid put jet fuel into that process, it completely accelerated it. I’ve never seen, I mean, we were working with carriers that were like, Hey, tomorrow we have to implement this entire new process. And it was crazy. And I think what’s good about that is that the insurance community can be slower to change. And that is I think well known and for lots of good reasons too, very risk averse and managing and thinking through risk and change is risky, but this really was a nice forcing function to get the industry to leap forward into the future.

(12:53):

And so I do feel like that was successful in general. I am sure there was definitely pitfalls and things, learnings that we had throughout the way, but it was materially successful. And so I think carriers are one of the other consequences of covid and just the lifecycle where insurance is at is there’s really this great resignation, if you will, in the field adjusting space. It’s hard to get new adjusters to come in. And so carriers are having to figure out, gosh, even if I didn’t want to downsize my workforce, my workforce is downsizing on its own. And so how do I implement strategies and technologies that will compensate for that? And this is where AI shines, and this is where a lot of these tools really help. It empowers people to do more with a lot less, and then it allows ’em to have better bedside manner with their insureds. And so I think this is something that covid really enabled in a strange way, but I really feel like it’s been a kind of pivotal moment where I don’t think the insurance community is going backwards. I think they’re just going to keep blasting forward with a lot of this transformation.

Patti Harman (14:06):

I totally agree. I’ve been covering the industry for at least the last 20 years, and I was absolutely astounded at how quickly all of the insurers, everybody, how they pivoted and how they were able to kind of change, well, we couldn’t do this, but we could do this, this, and this. And they just went on without missing a beat. And as I’ve spoken to different companies, well, how did you do this? And they literally did it. You’re right. They did it in a matter of days, weeks at the most because they had to adjust their claims process and how they handle customer service and everything else. And now with this new hybrid work environment, it really does lend itself to utilizing technology that much more,

Garret Gray (14:48):

And carriers are well positioned to do this because like I said, they were already going down this path, and so they already kind of understood what was possible. It was about kicking the tires and testing and doing proof of concepts. And so I think they had a good idea of like, okay, we know what to do. We just now have to do it. There’s something that’s actually forcing us to do it. And so a lot of the things that we were working on with clients just got greenlit and said, go, we have to figure this out right away.

Patti Harman (15:18):

Which must’ve been really exciting for all the different technology companies. Like, okay, good. I don’t have to wait for all of this. So can you give our audience some examples of how companies are incorporating AI technology then in new ways? I mean, we’ve been talking about the fact that they’re doing it. What’s it look like for them?

Garret Gray (15:34):

Yeah, so think about something as simple as when an adjuster or restoration contractor goes into a property that’s had a loss and they have to actually diagram that loss or the property itself and identify where all of the damages. Now you’ve been around for a long time. I have the ways people used to put on sketch paper and they would then take a picture of it, they would or scan it in. People got out measuring tapes and there was lasers. There was all different ways that we’ve been trying to diagram these properties. But now a completely AI powered process where a technician or an adjuster can walk through a house and just move their phone around and it completely diagrams it in a materially accurate way. Very, I think it’s easy to go and try to attack some of the accuracy. No one ever attacks the accuracy of the manual process, which is actually very inaccurate.

(16:39):

And so getting the to 97% accuracy is actually really, really great and better than the human manual process. And so think about just that alone has really transformed the way that carriers and contractors can operate speed up the process. And really what I like the most about what this technology does is it really frees up the people who are dealing with people who have just gone through a really traumatic and probably one of the more difficult things in their lives. It gives them the ability to be more human and to let the AI do the computer work and let them do the bedside manner. And I think for those who are embracing that and the ones that we see, we see the impact pretty materially.

Patti Harman (17:27):

That’s a really great way to describe one of the major benefits of this. We hear about, oh, it’ll free you up to do this, that, or whatever. But that’s just a really great explanation that along with that I think comes training and re-skilling workers so that they can effectively use AI safely. What are some of the ways that carriers are using AI to improve the insurance process internally for their employees? And then how important is employee buy-in into the implementation of these different AI technologies?

Garret Gray (18:01):

I think the one thing that employees have to get over, and usually I get asked this question in all of these, I know you’re probably going to go here, I’m going to head you off here. Is AI going to replace my job? And so what really happens is that people can buck the technology because they’re afraid it’s just going to replace them and they don’t want to be replaced. None of us do. And I think the thing that people need to hear is AI is not going to replace your job, but failure to embrace it might mean you’re replaced. And that’s just pure, not because AI is doing it, because somebody who understands AI or who has been using it and knows how to leverage it will outcompete you. And so it’s not something to be afraid of. It’s something to embrace. On top of the stuff that we work on for carriers, there’s things like a Microsoft copilot, and if you’re using Microsoft Copilot yet we are at CoreLogic, it automates things like you’re in a meeting and it can summarize notes for you and say, Hey, here’s a list of your notes. It can take an email thread and that might be 30 people going back and forth on something and just give you the, here’s a summary and tells you exactly what you need to know of those emails and can save you a ton of time. And so there are so many different ways that you can embrace ai, creating PowerPoint presentations, et cetera, and it’s just meant to speed people up, give them a starting place, and then let them use their expertise that they have to hone the 20% that needs to be honed.

Patti Harman (19:37):

I remember as a reporter, I’ve always had to do a lot of research and that sort of thing, and so using even Google being able to double check and fact check and all of that sort of research that would’ve taken hours and hours before I can do it in minutes without leaving my computer or my office. So if you look at AI in that respect, it really has so many different benefits and possibilities for us. We talked about how it affects businesses. How is the increased use of AI affecting the front facing customer service aspects of insurance?

Garret Gray (20:17):

So a couple ways that are obvious in some ways that are not obvious. So the obvious ways is now policyholders are getting more choice upfront. So maybe in terms of serving up who can respond to their loss in the dispatching process, the scheduling process, we’re really using AI to streamline how people have experts and adjusters and supply chain get to their property so that they get a fast response that is convenient for them. There’s AI driven eef and OL solutions that we’re working on and others are so that if you need a report, a loss that guides you through the process, gets the information, gets you, can you get settled right away or is there a way to verify that the weather, think about you make a hail claim, and we can very quickly by using AI in our weather forensics, go through and go, yep, there was a damaging hail in your area.

(21:21):

And so yep, we can just quickly say, yes, you have a covered loss, go get it taken care of. And so what I think you’re going to feel as a consumer is faster, better answers. And that’s really the key benefit is I don’t have to wait for a coverage decision for weeks. I can instantly know, okay, I’m good. I have peace of mind and I can get my property back to pre-loss conditions. So I think that’s one way that they fill it directly. The indirect is back to, I think carriers are really seeing this as a way that they can bring a more human element to their process, which might sound funny, talking about AI bringing a human element, but again, it’s the AI freeing people up to be more human and to automate a lot of what they do that can make them feel too busy to engage properly with an insured.

Patti Harman (22:11):

I think it helps make the process a little bit more transparent because people really know this is where it is in the process, this is how much long, this is who I need to reach out to ask certain questions if I’m not sure about something. And it just makes it a little bit easier for them on so many different fronts. Are there certain areas within the insurance ecosystem where you think the use of AI will really make a significant difference?

Garret Gray (22:37):

I think everywhere. I don’t think there’s any exception here from every lifecycle that we are involved with with the carrier, which is basically all of them, every single one is being impacted by ai, even in terms of how they figure out which policyholders are the ones that they want to go after in terms of ensuring, in terms of building their portfolios and where their policies are going to be enforced. Every part of the process is being impacted by ai, whether people know it or not. And I think that’s true just in our personal lives now. Most people don’t realize if you’re engaging with Siri, you’re engaging with AI. If you are, like you said, getting GPS directions and it’s telling you how fast you’re going to get there, and it’s actually monitoring traffic, that is an AI driven process that’s helping you get from point A to point B and how we live, work, and play AI is impacting it every day. And again, I’m an optimist, so I think this is going to be a positive thing. I don’t think we’re going to be hiding from the machines coming after us. I think we’re going to solve all sorts of health challenges and longevity challenges and maybe even economic challenges. And so I think insurance is one area that’s heavily being impacted, but as people who are just living in this broader world, we’re all being impacted.

Patti Harman (24:10):

Yes, we are without a doubt. We’ve talked a little bit about how workers are using it. The talent shortage though has been a major concern for the industry for years. Do you think that the use of AI will help augment some of those losses? I’m thinking in terms of institutional knowledge or even maintaining high levels of customer service for the industry.

Garret Gray (24:33):

Yeah, this is actually a really key benefit to AI because not only is there a talent shortage, but the tenure of that talent and the experience, the talent that exists is extremely new. We have got people who have not been in their position very long, don’t have, and especially in property. Property is very complex. It’s not a car that you can look up a VIN number and understand, okay, here are all the parts that were on this car. This is very different. And it’s hard to just come in and learn it. It’s a very difficult thing. So what AI gives you the ability to do is to take all the knowledge that you have, all the secret sauce and everything that you know about how to best settle a claim or to best underwrite a claim and to build it into a machine so that those who have less experience can get better insights to make better decisions.

(25:23):

And so yes, will it automate more things so that you can do more with less people, which again, isn’t going after anyone’s job, it’s just dealing with a world where we just don’t have as many people who are signing up for these sorts of jobs. So yes, it will do that, but it will also help fill in the knowledge gap because we have people who are been in the job for less amount of time than we’ve experienced in the past. A lot of adjusters are retiring today, A lot of contractors are retiring, they’re passing on to a new generation that doesn’t have the experience. AI can help bring that experience to bear across the entire platform.

Patti Harman (26:03):

That’ll be really important. I was thinking in terms of construction, because a lot of people don’t understand what goes into building a building or what are the different aspects of it, or how does water drain always to the lowest place? So if you have a loss on the third floor, yes, the water’s going to drain all the way to the basement or whatever it is. So I can see where AI would definitely be very helpful. How do you think technology is going to continue to change the insurance process in the coming year or so? I used to ask people two or three years out, but technology is changing so fast, so I’m thinking maybe in just the next 12 to 18 months.

Garret Gray (26:40):

Yeah. Look, I think that one way that there’s going to be a big change, at least that we’re a part of and are driving is the time that it takes to get a resource out to a loss can really impact the overall severity of that loss. So for example, if I actually, I have a good example. I had a loss recently, I won’t get into the details. It was like black water that flew out from a toilet. Our kids did something they shouldn’t have done anyways, so we had water come up, it was dirty water, it was all over kind of the bottom floor. Thank God it was on the bottom floor. And I was able to call, make a claim. I was able to get a contractor out really quickly, and they were able to save a lot of the damage and restore it without having to rip it out purely because of how fast they responded.

(27:37):

And so what is that process though in the past has been really hard. Filing the claim was hard. Then getting the adjuster to help you assign a restoration contractor took time. We’ve now automated that process all the way from the point of FNOL to figuring out who is the closest, most best provider that can go out to that loss and let’s get ’em out there fast. And so those things, that’s just one example of where AI can really impact the process and not just give the policy holder a better experience, but also it helps with severity because the faster we respond and with the right type of equipment, the less damage is done to properties.

Patti Harman (28:24):

And that’s what a lot of people don’t realize that time really is of the essence for some of these losses, and it just makes it worse as time goes along. Are there uses of AI that you find particularly compelling for the insurance industry that just really kind of excites you about, Ooh, this is going to be really cool and wait until you see what happens in 12 months or something like that?

Garret Gray (28:46):

Yeah, I think image analytics is what I’m mostly excited about because if you think about what I was talking about where you could use your phone and get a diagram of the room instantly, that’s a form of image analytics, the ability to detect the condition or the quality of materials, things that you might have had to send out to somebody before to get a, what is the quality of that particular material? Being able to do that faster is coming. So I think both from the interior and the exterior, the ability to look through an image, compare it to a past image, do a detailed analysis on the condition and the quality and what happened and what needs to happen next, all of that just really takes what was very manual and required a lot of elbow grease to get done and just speeds it up. So we’re on the precipice of that, right? That is, you’re seeing it come in through all of our products already, but it’s about to kind of coalesce and we’re getting close to that kind of image to scope process that we think can be completely automated.

Patti Harman (30:02):

Wow, that really is exciting. I can understand why that would be something that you would really kind of flag. Are there aspects of using AI in the insurance industry that maybe also cause you some concern?

Garret Gray (30:16):

Yeah, I think the biggest concern is that there might be some providers out there that don’t have the right data sets that are sort of rushing out there with models that will give bad outcomes and bad predictions that then maybe taint carriers from wanting to embrace other models. So I think because if you have an experience where it didn’t work right, and where you got bad outcomes, you might be less likely to try another platform. I think it’s really important for carriers to look at who has the best data for what they’re trying to do, who knows property the best, and make sure that they have the underlying data that trains these models to get to better outcomes. And so my biggest concern is there’s a ton of people running after it. Not all of ’em have access to the same level of data, and we don’t want erroneous results out there. Kind of tainting the overall idea of using AI because AI is so transformational that if we want people to be embracing it and making their processes more efficient,

Patti Harman (31:29):

Are there certain risks for challenges then that companies should be monitoring as part of their implementation process?

Garret Gray (31:36):

Yeah, I think it’s really important as you implement and you’re sitting alongside these AI driven processes that you have humans verifying what’s happening. So the idea is that it can speed up what they’re doing, but you don’t want to hop to an AI trying to make a decision on its own, right? You want the AI to enable the person who is making the decision to do it faster and with better accuracy and quality. But you don’t want to try to jump too fast to like, Hey, we’re going to just take an image and let it tell us what the estimate should be like. That’s not happening yet. You still need to have a, and I think for a long time we’re going to have to have experts looking at what the output is and tweaking it and kind of giving feedback to the AI that, yeah, okay, you got it right this time. And so I think anyone who jumps too fast and to allowing the AI to make decisions first as opposed to enabling humans to make those decisions will struggle.

Patti Harman (32:36):

That’s an important differentiation in terms of we’re still training the AI how to do some of these things. We can’t just assume that, oh yeah, here, let me just throw you all of this stuff and let’s see what you can do with it.

Garret Gray (32:47):

And if you think about it, they’re like a new employee in a weird way. They have massive amounts of access to data and hopefully it’s good data, but they still need feedback. It needs feedback, it needs to be trained. And that is an ongoing process. And so it’s important for you to have humans that you trust, validating the outcomes, making sure that it’s what you expect, and then starting to implement it.

Patti Harman (33:15):

So true. We have covered a lot over the last couple of minutes, but I’m wondering, is there anything that I haven’t asked you about the adoption of AI across the insurance industry that you want our listeners to know?

Garret Gray (33:29):

Listen, I think we’ve hit on this, but I’ll just reemphasize. This is not something to be scared of. It is something to treat seriously and to really think through and be diligent about how we implement these tools, how we build these tools and the effects that they’re going to have. But for me, the possibilities aren’t limited, and it’s really exciting to see our community, our insurance community, embracing these sorts of tools post covid in a way that they are. And I think that the benefit for policy holders, carriers and everyone in the supply chain is going to be immense. And it’s just everyone getting together, bringing multiple stakeholders together. There’s not one company that can do it on their own. And having really good open, honest communication around how these things work, how they integrate together, the type of decisions they can make, kind of decisions they can’t make, and how do we enable people to be their best at work. I think that’s the secret sauce here. Bringing people together, have the conversations and really work on implementing, validating, and moving on to the next thing that we can automate.

Patti Harman (34:45):

Wow. Thank you so much, Garrett, for sharing your insights with our audience. This is such a fascinating conversation. I learned so much when I do these interviews, so it’s just such a privilege for me too.

Garret Gray (34:58):

Oh, great. I enjoyed it.

Patti Harman (35:00):

Thank you for listening to the Dig in podcast. I produced this episode with audio production by Kelly Maloney. Special thanks this week to Garrett Gray of CoreLogic for joining us. If you’d like to learn more about the use of AI and other InsureTech in the industry, join us at the Digging Conference in Boca Raton, Florida on June 27th and 28th where we’ll be showcasing a variety of new technologies and discussing technologies, impact on claims, customer service, and data security. Please rate us, review us, and subscribe to our content at www.digin.com/subscribe From Digital Insurance, I’m Patti Harman, and thank you for listening.