AI Wisdom Ep. 7: Ferreting out Auto-Insurance Fraud with AI

Digital Transformation - March 4 2020

On this episode of the “AI Wisdom – Talking Innovation in Insurance” podcast, host Ron Glozman speaks with Eugene Greenberg, CEO and Founder of Click-Ins Ltd., about how insurance companies are using intelligent technologies to automatically detect and predict auto-insurance fraud. Ron and Eugene also discuss how self-driving vehicles, ride- and car-sharing are disrupting traditional insurance business models and share insights from the trenches on their insurtech journey. Click the link to listen or read the full transcript below.

 

Full Transcript

Ron: Hello, and welcome to “AI Wisdom - Talking Innovation in Insurance.” On this podcast, we talk to business and insurtech leaders about how artificial intelligence is transforming the way we buy and sell insurance. I’m your host Ron Glozman, founder and CEO of Chisel AI, and a strong believer in the power of AI to help people work smart and enrich their lives. So, let’s get into it.

This week, we’re joined by Eugene Greenberg, CEO and founder of Click-Ins Ltd. We’re going to be talking a lot about reducing the risk of costly, fraudulent auto insurance claims. It’s now easier than ever for insurance companies to combat this with intelligent technologies like Click-Ins’ DamagePrint™. With new business models like ride-sharing, car sharing and self-driving vehicles all disrupting traditional auto insurance models, both on the commercial and personal side, companies are now looking to embrace new approaches and new technologies to detect and predict the fraud that we all know is abundant in today’s environment. Eugene, it’s a pleasure to have you with us. Can you please introduce yourself?

Eugene: First of all, my friend, I’m so happy to be here with you. I’m Eugene Greenberg. I’ve spent almost 30 years in insurance. That’s due, I guess, to my strong domain expertise in this industry. We established Click-Ins Ltd. in 2014. Since then, we are tackling the biggest problem in insurance industry, insurance fraud.

Ron: Love it. And can you share a little bit about this? I’ve seen some of your work, and I know you’re currently participating in the Zurich Innovation World Championship. Why don’t you give everybody a 30-second highlight of some of the results you guys have been able to show to date.

Eugene: First of all, I’m so excited! In a couple of days, I’m flying to Chicago to pitch Click-Ins in front of top C-level management from Zurich and, hopefully, we will win this regional final in the U.S. Since last year, we had several breakthroughs in technology, and we won the BASF Damage Recognition Challenge. BASF is the biggest German company that produces paints for car manufacturers. And, hopefully, our technology will be implemented in their product. And the reason that they chose us is it appeared that we are the only company in the world that not only can recognize all kinds of damages on the surface on the car, but we can very precisely measure them – we can tell you that this particular scratch is 5 centimeters.

And the core technology that turns that particular scratch into a DamagePrint is very similar to a fingerprint: that scratch becomes a very unique identifier of that particular damage. And that’s how we tackle insurance fraud because each time that that particular scratch appears in the picture in any type of combination of damages or build-up damage or a combination of damages, we can recognize that damage as a previous one and alert an insurance company of possible fraud.

Ron: Love it. And I know, speaking from experience, as somebody who lives in Canada, in Ontario, which has the highest or one of the highest rates of auto fraud in Canada, they did an investigation, I was just looking up the details. Aviva, which is one of the major insurance companies here in Canada, did an estimate and found that 57% of their payouts are actually fraudulent. And that’s just crazy. That makes it more than half.

Eugene: The crazy thing in that report is not 57% of payouts. It’s that Aviva is actually the first company that ever proved empirically that 9 out of 10 motor insurance claims involved fraud. And when we’re speaking about fraud, I would like to explain something. The fraud divides into two major parts. You have the criminal activity, okay? So, it’s big cheques. It’s always fraud rings. There are a lot of people involved. You have body damage to a car. Then you have a whiplash, all kinds of personal damages. So, insurance companies can tackle that type of fraud because it’s a big check. An insurance company can allow themselves to investigate it. They can send that particular case to SIU – Special Investigation Unit. I would estimate that this type of fraud is maybe 25% or 30%. That’s exactly what you see in all the statistics because when you see those fraud statistics, let’s say in U.S. – like $120 billion in fraud reported by the FBI – only that type of fraud is investigated.  

The rest of the fraud, 70%, we call it soft fraud or friendly fraud. This type of fraud is not performed by criminals. This type of fraud is performed by usual citizens, you know. They’re not doing it for a living. They’re doing it just to fix the car. And the most sophisticated scheme, it’s build-up damage or bumped up damage, okay? So, imagine you have a small damage on your car under a deductible. You go to a body shop, the body shop bumps it up and, voila, you submit the claim to your insurance company, and get the money. Or, you have several damages, and each of these damages is under a deductible. You go to a body shop, bump these damages all together, voila, you have a claim. You submit the claim to the insurance company, get the money, fix the car. And you can even make up all the spare dollars, okay? That type of fraud goes under radar. Nobody investigates it due to a huge, enormous number of such claims, and these claims are mid-size and small-size claims. So, insurance company cannot afford to investigate these claims because they’re only going to lose more money.

Aviva is actually the first company that ever proved empirically that 9 out of 10 motor insurance claims involved fraud.

So, when we started to tackle that type of problem, we understood that the only approach that will cut off this engagement is to mark each and every damage into a unique identifier and then, by matching it to a repository or to baseline, find the fraudulent engagements. And that’s what our technology does.

Ron: Love it. And you would never imagine when you talk more about the deductible that the average citizen – even unknowingly sometimes – can fall prey to that. Even though they’re the instigator, they are unknowingly maybe participating, and so I can imagine how difficult it must be to actually conduct an investigation. And so, I can say for sure that I see the need for this. So, I’d love to hear… There are a lot of new, cool technologies on the market. For example, Uber, Lyft, and some self-driving tech are starting to come out. How do you see all of these new technologies impacting the way the insurance industry works and also the personal insurance industry because there is a little bit of a difference, as you know, between the two?

Eugene: Okay, first of all, you always need to see the business side. The business side is very simple. None of the insurance companies anywhere in the world makes money on motor insurance and car insurance, okay? That’s it. The loss ratio is above 100%, okay? So, maybe one year, they can make a couple of bucks, but if we take the average over five years, all over the world, the insurance companies are suffering. The biggest problem is claim costs and fraud costs, okay? Of course, you have acquisition costs.

So, what I see is that the insurance is coming through a lot of changes. And companies like Uber and Lyft, they bring their cut to those changes.

I don’t believe that the insurance industry will be disrupted that easily, but it will be innovated, for sure.

There are more and more technologies I see today for claims and fraud engagements. Insurance companies take these technologies and implement them as fully automated processes because they want to avoid human mistakes, human fraud, human interference. The same goes for the rentals and car-sharing. And Uber, as far as I know, is entering the car-sharing business, and tomorrow they will need certain automated procedure for damage recognition and damage matching to avoid human errors and human inspections, okay, as well as the cost of human inspections.

So, all these companies, of course, bring their own parts to what we call innovation in the most dinosaur industry, insurance. But, I guess, in a few years, we will see a lot of changes because the insurance companies want to make money and they don’t today. And that’s the biggest problem. And then, in the end, a customer is paying a more expensive premium, and the customer didn’t get paid in claims because the insurance company doesn’t make enough money. So, as I said, in a few years, we will see much more innovation in the industry.

Ron: As you think about self-driving cars, obviously, you know, we still haven’t seen them. Maybe we will see them this decade. Who knows? But one of the things I’ve heard, especially with self-driving cars, is this question, which is, who pays for the insurance and who is liable? Is it the software provider? Let’s say it’s a Toyota manufactured car, and the AI comes from Tesla, let’s just say they hypothetically license that out, who is paying the insurance premium? Or if it’s Ford manufactured and Ford AI, is Ford on the hook? Is it the owner of the car that’s renting that out for ride-sharing who is on the hook? Where would you think – and obviously, this is a couple years out, but if you had to take an educated guess – the liability lies?

Eugene: Frankly, you know, what we first heard about autonomous, the predictions said we had a couple of years to go, like, we were on the verge of seeing a lot of autonomous cars in a few years. Today, the industry talks about 10 to 15 years. Not that fast, okay? Frankly, nobody knows exactly who will be issuing the policy for autonomous cars and how it will work. I guess the best way is when you buy the autonomous car, or you rent the autonomous car, the price will be included in the full package of full coverage including liability and whatnot. That’s what comes to me right now. But, you know, I’m not coming from the actuarial part of insurance, so I’m not sure how the insurance company will calculate the risks of autonomous cars. Currently, nobody knows. This is something like a black box. Nobody knows yet, so we will see how it will develop.

Ron: For sure. And I think it’s an interesting point that you raise, which is, you know, people were very optimistic, and optimism is always good, but you need realism, especially as you get more information. And so, it’s interesting that now in your conversations, you are hearing 15 to 20 years. That’s a very interesting point. So, on another topic… You know, AI has been a big game-changer for insurance as a whole, not just for auto companies individually. When we talk about AI and automating specifically vehicle damage detection, and you talk about the loss ratio being above 100%, how do you foresee them being able to deal with some of this soft and hard fraud? And where do you think, optimistically, they could get their loss ratios down to?

Eugene: Let me divide that question in two parts: the AI issue and the loss ratio issue. First of all, let’s talk about AI. AI, it’s a very beautiful buzzword, you know. In our company, we don’t like to mention the use of AI too much. We use a mimic intelligence, not artificial intelligence, because, you know, it’s not really artificial intelligence. It is something that is closer to mimic intelligence. But AI has its own limits, okay? It has very good abilities for certain tasks. At Click-Ins, we are not using only AI or deep learning. We’re combining deep learning with 3D modeling and CAD and a bunch of engineering disciplines that, in the end, give us a very high precision in measurements, high precision in recognizing parts or damages. Using only AI is not going to give you a comprehensive solution. And don’t forget about the AI adversarial quadrant, okay? Because if you have a set of a zillion pictures, that’s enough that a couple of pictures will be biased with something. And that’s it. You have an adversarial quadrant. So, we use a slightly different approach. We’re combining AI and engineering disciplines to get more precise results.

So, the second part of your question: how to reduce loss ratios? That’s exactly what we are trying to do, to help an insurance company reduce loss ratios by automating the whole system of claim settlement and fraud engagement by cutting off any human interference in that process and allowing the so-called computer to decide what’s right and what’s wrong. And, as I mentioned before, the money of an insurance company lays in the claim cost and in the fraud. So, we’re using our techniques, including AI, to help an insurance company cut off these expenses by automating all of the procedure. That’s the only way.

Ron: I love that. And I have actually talked about this before on this podcast. You know, AI is a wonderful technology, but it has its downsides, and you mentioned some of them. We’re also big believers. I mean, our whole company even has the name AI in it, and our core products are based on it. But you only use AI as little as you can because, to your point, if you can have higher precision or certainty using more traditional methods, you should always do that. There are some cases when AI is necessary, and AI is, for sure, the right answer. But we’ve done a lot of talking on this podcast about innovation for the sake of innovation and how that is not the right approach. You should always do innovation for the sake of a business case rather than just doing it to say, “Here is something cool I can build.”

Having said that, I’d love to switch sides a little bit and talk a bit about what it takes to build an Insurtech. I read a study recently that said that, in 2014, there were approximately 3,600 insurtechs entering the insurance market with new business models and technologies. As someone who has been in the industry for 30-plus years, and you’ve been working on this since 2014, what advice would you give to an insurtech that’s listening right now?

Eugene: First of all, I read a recent article from Zurich’s Head of Innovation and he said a beautiful thing. We, a couple of years ago, were blinded by technology, and we were trying to implement everything, blockchain and this and that. And then we understood very quickly that in 90% of instances, there’s not a business case. That’s exactly what I believe. First of all, to be an insurtech, you have to have very strong domain expertise in insurance. Insurance is not GetTaxi, it’s not WeWork. It is not something that’s simply or easily disrupted or innovated, you know? If I’m taking you back to 2014 or 2015, a lot of people were saying, “We will disrupt this industry.” Today, we don’t hear that. Because now everybody understands you cannot disrupt the oldest industry in the world, a $5-trillion industry like insurance, okay?

So, first of all, my advice is you have to have strong domain expertise and you have to be a founder with domain expertise. Second, don’t ever try to sell algorithm to an insurance company. They don’t buy it. They buy a solution to their problem. The third piece of advice, and it’s very important, is an insurance company is a huge institution with a lot of departments and sometimes, not sometimes, always, one department can be in so-called war with another department. So, you need to understand who to sell your solution to. You need to understand, “Who is your customer within the insurance company?” If you’re selling, for example, an underwriting solution, you don’t need to talk to claims or marketing, okay? You need to talk to the underwriting department. And if you’re settling claims for an insurance company, your customer is the chief claims officer because his bonuses depend on how low the loss ratio is in that company. So, if you can bring that to him, he’s going to be the best customer you ever had.

The next thing is, you know, I’ve heard a lot of young guys, a lot of insurtechs, coming to insurance companies and saying, “Listen, you buy our solution, and you can fire half of your company!” And that’s it. That’s the last day the insurance company will be talking to these insurtechs: “Don’t call us, we’ll call you.” Because nobody likes to fire people. Don’t forget about the unions. They’re very strong in the insurance industry.

So, the key point is to bring more money to an insurance company. You don’t want to interfere in their business procedure because once you starting interfering, that’s it, okay? It will take centuries to get inside.

Try to build a parallel business process with the insurance company. Try not to invade their business process. And that’s the best solution, that’s the best advice. Be confident. Be a domain expert in what you do. Try to not invade their business process and bring them more money than they do today.

Ron: That was so well put. I think we’re going to have to clip that. That was so perfect. So, we’re going to take a 20-second break to tell you where you can find out more information and insights about insurance innovation. We’ll be right back.

[If you liked this episode of AI Wisdom, subscribe to our blog, Writing the Future: AI in Commercial Insurance at www.chisel.ai/blog for feature articles, interviews, opinions, and more.]

We’re back with our featured guest, Eugene Greenberg. Let’s jump right into the next question. On the flip side, a lot of these insurance companies have what we call analysis paralysis. You know, they want to run POCs forever. They want to benchmark six solutions at once. They want you to do it for free. All of these things that, as an insurtech, you often come across. What advice would you give to somebody, you know, on the flip side who is working with these solutions so that they make the right technology investments?

Eugene: There has to be commitment. If someone tells you, “Okay, let’s do the POC for free,” their commitment will be so low, okay? And they won’t cooperate, and then you will be mailing back and forth, and everybody passes through this, okay?

We have a rule in our company. We don’t do pilots or POCs for free. That’s it. We value our time, our people, our team, so why would we do something for free?

Let’s be frank: what an insurance company is afraid of is that you will take their money and do your R&D using their money. This is something different. When we’re talking about the POC or pilot, we always take money, and we’re not getting this money for profit, it’s only for covering our expenses. And I’m not doing R&D with that money. I’m just showing them our capabilities on their data set, and they could be the first to implement my solution and to make more money. So, you can negotiate with them. For the first 100,000 documents or 10,000 claims, for example you can give them a 50% or 70% discount. But I wouldn’t do something for free because each job has to be paid.

To prove that your technology is a good investment for an insurance company, it’s very simple and it goes back to your previous question: Are you enough of a domain expert in that industry? Usually, in insurance, it’s the same people who work within the industry for decades, okay, decades. And they know each, and they know every corner of the business, which someone without expertise doesn’t. You need to be so expert, so confident to prove to them that you are the real deal, that you are exactly what they’ve been looking for over the past five years, okay? That’s the main point. It always comes down to domain expertise. You know, Oscar Wilde said, I think it was Oscar Wilde, that people have to do only something that they are the best at, okay? That’s it. I don’t believe that someone who has been working in a different industry will be that successful in insurance. It’s a tricky industry. So, again, domain expertise is key.

Ron: I’m with you. I’ve had several conversations with people both professionally and in my personal life that there’s two ways to live life. And I’m not certain to say that one is better than the other, but you can be a, you know, a jack of all trades and a master of none, or you can be a master of one trade and a jack of none. That’s just the difference between going wide and going deep and, you know, some people go wide, some people go deep. I’m with you that, in the insurance industry context, depth is much more relevant than breadth.

So, in your opinion, what is the risk that insurance companies face by not transforming the way they conduct business? We talked a little earlier about, you know, people used to talk about disrupting the industry, and they have moved away from that because we all understand that disruption might not be the right word. But on the flip side, insurance companies that don’t change will be left behind. What does that mean to you?

Eugene: You know, it’s a tricky question. First of all, I don’t believe that Allstate or Liberty Mutual, if they do not implement all the technologies in the world, will be left behind. No, you need to understand insurance – it’s a lot of money. It’s a lot of money in pensions and in insurance, life insurance coverages. It’s not that simple. It’s not like GetTaxi, you know? In these industries, you’re coming with a new model, and that’s it. When was the last time you bought a ticket through your travel agent? I’m buying through Expedia, you know. That’s how new technologies can kill a business.

The insurance business is something else entirely. The insurance business, it’s a very regulated, heavily regulated business. It’s not going to disappear very quickly. Look at the 2008 financial crisis: Nobody disappeared, not one insurance company. Even banks went bankrupt, okay? But the insurance companies including AIG, they stayed, and the state helped them. That’s the point.

But, if we’re talking about making more money, of course, they will be behind if they don’t implement new technologies. And so, they must do it. Not the way we want, not at the speed we want, but definitely they must embrace new technologies. I don’t think they will disappear, and I don’t think that newcomers, new companies will just, you know, cut big chunks out of their business. I’m not confident in that.

But what I see is they’re innovating. All of the companies, all of the big ones in the States, Canada and Europe, especially, are innovating. Look at Zurich, okay? Zurich is hosting the global Zurich Innovation Challenge for the second year in a row. They’re trying to get the best of the best of the best. Munich Re is doing it too with Munich Re Digital Partners. They’re always looking for technologies and not only for technologies but new business models too. They’re looking for technologies that can change the face of industry. That’s what’s important. And Munich Re is the biggest reinsurer, and the more insurance companies lose money, the more reinsurers lose money as well. So, this is all in.

So, I believe they will be staying. They will be innovating in their way, and I don’t see that much changing in the market, if we’re talking about the big companies. You know, in each country, the insurance business is very concentrated. Look at the States. In the U.S., there are 5,800 companies, okay? And 30 companies have 40% of the market. That’s it. And in every country, it’s the same. Take Japan, 3 companies, 30%, 30%, 30%, and that’s it. So, let’s see how they will be implementing our solutions, my friend.

Ron: For sure. But I think you made a really good point that I just want to highlight and I think it was very well put, which is they’re going to be here for sure, like, there’s enough complexity and money and history there that they’ll be around. What they’ll be missing out on is the net new revenue being driven, and I think that is a very good way to put it, which is, they’re not missing out on anything in terms of what they have today, but they’re reducing their share of the new pie being created through the efficiencies that these technologies bring.

Eugene: Correct, correct. They will be keeping their share of the market, but you can bring them more money on the same share of the market. I would say it like that.

Ron: Let’s wrap up. We like to do two questions, one-minute answers each. It’s the bullet round. If you’re ready, let’s do this! How will AI and other technologies impact the insurance industry in 2020 and beyond? And so, this is the question: I’m going to say, in five years, where do you think insurance will have gone? In one minute. 

Eugene: In one year, I don’t see AI changing much in an insurance company because you need to implement it, and you need to implement solutions. And implementing solutions, that’s what takes a lot of money, okay? Not the AI itself. The implementing of the product, implementing of the technology, that’s what takes time, okay? So, in one year, I don’t see that much changes. In five years, I guess there will be a lot of changes. We will see much more automated claims and fraud engagements. We will see all OCR documentation fully automated. You will see no human interference. We will see more direct marketing without any agency. It will be bots. Something like that.

Ron: Yeah. I’m with you. And so, to wrap up, if there was one piece of wisdom that you wanted to give to our listeners in any context, what would be that one piece of wisdom?

Eugene: You know, I’ve been involved in sports for 40 years. For half of these 40 years, some 20 years, I was in professional sports. You know, a lot of people say, “It’s important to participate.” No, it’s not important. It’s important that there is one medal: There’s the gold one, and then all the rest. So, in starting a business, it’s all about the gold medal. I call it zero optionality. When you decided to open a startup and go with it, you should have zero optionality. You should never think, “If I’m not going to succeed, I’m going to go work as a software engineer in a, I don’t know, in a different company.” That’s what will kill your business, okay? You have to have zero optionality. Go with God, okay? Believe in yourself and never look back. And the more the venture capitalists say “no,” it has to make you stronger. Don’t be biased by their “no’s, go in there and you will get the “yes.” That’s my advice.

Ron: There will be people listening to this who needed to hear that, so thank you for that. Thank you for taking the time. If anybody wants to reach out, please follow Click-Ins on Twitter, LinkedIn. Where can they find you, Eugene, and where can they find your company?

Eugene: They can find us in Twitter and Facebook and LinkedIn, click-ins.com. Please send us any questions you want. Talk to us. If you need advice, send us an email.

Ron: Awesome. Thank you, Eugene, and thank you, everyone. Have a great rest of your day.

Eugene: Thank you, my friend. It was a pleasure speaking to you.

Ron: That’s a wrap for this episode of “AI Wisdom” hosted by Chisel AI and me, Ron Glozman. Thanks for listening.

If you like our podcast and want to hear more, check us out at www.chisel.ai or tune in and subscribe wherever you get your podcasts: SoundCloudiTunes or Stitcher

Join us next time for more expert insights and straight talk on how AI and insurtech innovations are transforming the insurance value chain. See you on the next episode!

 

Browse different topics

Recent Posts