When he spoke at Business Insider IGNITION: Transportation in 2019, Gary Hallgren, President of Arity, discussed the importance of collecting and analyzing information about how human drivers interact with other drivers, as well as how humans interact with self-driving cars.
With millions of miles of driving data plus access to claims data, Arity can “identify and quantify risky driving unlike anybody else,” accurately predicting what kind of driving will cause accidents, and how much they will cost. The full Business Insider article can be found here.
Gary Hallgren, President of Arity, discusses the company’s innovative approach to data analytics in the context of driving behavior and insurance. Arity, a spin-off from Allstate, leverages extensive driving data and claims information to identify and quantify risky driving behaviors. This data-driven approach allows for more accurate insurance pricing and risk assessment. Hallgren explains how Arity collects various types of driving data, including speed, acceleration, and phone usage, and integrates it with claims data to develop predictive models. He also addresses privacy concerns, emphasizing the importance of transparency and regulation compliance. The conversation touches on the future of autonomous vehicles and their impact on the insurance industry, with Hallgren predicting a gradual evolution rather than a rapid transformation. He highlights the potential for Arity’s analytics to enhance safety and efficiency in transportation, both in personal and commercial contexts. The discussion concludes with insights into the competitive landscape and the broader implications of AI and data collection in the mobility sector.
Matt Weinberger:
Oh, there’s our slide. Hi, I’m Matt Weinberger—
Gary Hallgren, Arity:
You didn’t forget my name, did you?
Matt Weinberger:
I did not.
Gary Hallgren, Arity:
That’s why you were looking—there’s a slide—
Matt Weinberger:
I was waiting for our slide to show up.
Gary Hallgren, Arity:
Oh, OK. Just checking.
Matt Weinberger:
This is—the comedian to my left is Gary Hallgren with Arity. I’m Matt Weinberger. I’m a senior editor with our San Francisco office. Arity, artificial intelligence data—how do you explain to people what you do at parties?
Gary Hallgren, Arity:
What do I do at parties?
Matt Weinberger:
Oh no, how do you explain what you do at—
Gary Hallgren, Arity:
Right. Because we were talking about the Star Wars thing earlier.
Matt Weinberger:
That’s true. We were.
Gary Hallgren, Arity:
So, it could go all sorts of directions.
Matt Weinberger:
Tickets on sale now.
Gary Hallgren, Arity:
Right. So, Arity is a data analytics company and the way I describe it is we can identify and quantify risky driving really—we think unlike anybody else in this day and age. So, you think about that person that’s driving in front of you and they seem to be reckless, they seem to be going fast, they seem to be looking at their phone, that blue glow that we all see. So, I think we can all identify these things. We grew out of the Allstate Corporation and have been able to collect so much driving data, but more importantly marry that together with actual claims data so that we actually know what is risky, and what isn’t risky, and what is going to truly lead to an accident, and maybe as importantly, truly what it’s going to cost to ensure that particular behavior.
Matt Weinberger:
That’s right. And then when you say you quantify that, what kind of data do you collect?
Gary Hallgren, Arity:
Yeah, the data—so, if you start—I’m going to start a little bit from what is insurance because it’s an interesting way to think about it. I was a technology person, not an insurance person. So, in my simple mind, you think of insurance companies doing something that is really hard for us to do and that’s to sell something that you don’t know what the cost is. You have to guess, and you have to make it up, but you can’t guess wrong, or your shareholders are upset, and you can’t overcharge because regulators get upset. So, you have to be very accurate at doing that. So, I like to think of it like a jar of marbles. You have a jar of marbles and there are all sorts of different colors, and as you’re looking at these individual colors, you learn over a course of time how you can price this group of drivers based upon that set of colors.
If somebody comes in and starts to weigh the marbles, and they’re weighing the marbles, and they’re taking—realizing that all the heavy ones are actually the good risk ones, and they take those and you’re left with the same color distribution, you start to lose, and you start to lose badly, and you lose over a period of years. And, actually, what Arity does is we bring a scale to the equation so that you’re now looking at risky driving in a different, totally different way. It’s interesting is that it’s an insurance thing—yes—but when you read the S-1s of the ridesharing companies going public, you see how important insurance is to them—with the self-driving, it’s is going to be that way as well. Anybody doing commercial trucking, it’s that way as well. So, everyone is trying to figure out this new way of doing it, and we have a scale in a way that no one’s been able to do that before. So, to the question of what data?
Matt Weinberger:
Yes.
Gary Hallgren, Arity:
Yeah, so, we collect data—a lot of it is just the driving data that exists today. So, it can be things around speed, it can be things acceleration. It’s interesting, like people say, well, you brake hard. And I looked it up when I joined Arity and said, Well hard braking—I was going to look in the dictionary, the encyclopedia, and the Bible, and there is no description of what a hard braking event is, but if you take billions of miles of driving data, and you put it together with thousands and thousands of claims, you start to understand what that is. So, from a device, we can get speed and acceleration. From a phone, I get speed, acceleration, plus, I also get, like, the barometer. I get if you’re using your phone—did you pick it up? How much did you pick it up? When did you pick it up? And we start getting into the OEM vehicles and now I understand that the windshield wipers are on, that the anti-lock brakes went—those types of data. So, it’s all grabbing that data, matching it with claims, and with that, you can start to do all sorts of interesting use cases.
Matt Weinberger:
That’s right. That’s right. But then, when you talk about how often you pick up your phone and all of that, how do you think about the privacy trade-offs when you collect that amount of data?
Gary Hallgren, Arity:
It’s interesting. The way that insurance today—everyone, for a long time, they’ve been trying to figure out how to assess this risk. So, one of the biggest variables that they use today is credit data. People think of it—your credit score. It’s not exactly that, but you could almost think it that way. The biggest proxy today, to set the price of what everyone pays, largely has to be related to their—is related to their credit information, which no one—doesn’t make any sense. So, the idea that you’re going to measure my driving and that driving is actually going to correspond to what I pay for my risk—that makes total sense. When you think about it from that insurance standpoint. When I think about it from ridesharing, for example, I want my shared mobility driver to have five stars because indeed they’re a good driver and not just that the car didn’t smell and they gave me a mint. So, I think there’s elements throughout all the different modes of transportation that are changing and people want to understand this risk. That’s really what’s going to push these new forms of transportation along.
Matt Weinberger:
That’s really interesting. And then to that point, you mentioned relationships with OEMs. I think about how you can buy a cheap 4K TV—it’s subsidized by partnerships with advertising providers. Will I be able to buy an Arity edition car in 10 years?
Gary Hallgren, Arity:
I don’t know if it’s an Arity edition car. The way that I—and I love the conversations on the autonomous vehicles earlier. Today, what we have is we have models and algorithms that help people understand how do humans interact with other humans. I’m convinced, as the autopilots of the world, the level two stuff starts happening, our models will start to be that this is how humans interface with cars that are kind of run by themselves. And the way that I usually describe it is that in the future state, there’s going to be Volvos and there’s going to be BMWs. And the BMW’s going to get on the road and realize that the Volvo cars—they’re so nice, they’re Scandinavian, I’m the ultimate driving machine, I’m going to pass them, and they’re going to break for me because that’s just the way the Swedes are, kind of thing. So, we will have models on how all the autonomous vehicles interface with each other, as well, because ultimately there still is going to be risk. It’ll be less. They’ll be less deaths. I think that’s a given. But all of these models will evolve over time and Arity will be the one to power those analytics.
Matt Weinberger:
Interesting. And then, you know, we think a lot about analytics and insurance, all that, in real life. Can you talk a little bit about how that data is being put to use today?
Gary Hallgren, Arity:
Yeah, so the insurance is—that’s one element of risk. I thought it was really interesting. I mentioned ride hailing earlier—just how much—it’s like almost insurance got sexy during that, when you look at the dollar amount that they’re spending in doing that. But, what we’re doing is we’re taking these analytics and we’re bringing them to other apps. So, we have a relationship with Life360. We do collision detection for them. So, we’re able to understand that. With weather companies, the idea that you should be leaving earlier—I understand your commute to and from work. You should be leaving earlier today because we know that hail is coming. Another example with those that have kids. My kid’s a pretty good driver, but when he’s with Billy, it’s like, Billy is kind of a nut, and you would’ve no way of knowing that unless you understood it. So, we’re taking a lot of these same analytics, going to where the data is, making transportation safer, smarter today, while we get to the future that is more safe and more autonomous.
Matt Weinberger:
I always knew that Billy was bad news.
Gary Hallgren, Arity:
Right. It was never you though. You’re not a bad guy.
Matt Weinberger:
Exactly. That’s right. That’s right. That’s right. So, talk to me a little bit about how you see the competitive space and who do you see as your competitors as you chase this?
Gary Hallgren, Arity:
When we think about it, we grew out of, grew out of the Allstate corporation. It’s a bit of a gift because we have access to all of that claims data. So, a car company, for example, like a General Motors with OnStar, for all these years, they should be able to develop algorithms that would know that if people drive this, they’re likely to airbag go off. What we have is the ability to understand this is more likely to file a claim, or this is what it’s going to cost, or this is who got hurt. So, just having that more data—we’re with one of the ridesharing companies here and they just looked at me and said, You have an unfair competitive advantage over anybody who doesn’t grow out of that side. So, there’s a group of tech companies that are doing what we’re doing. Our goal is to go to where the data is with other app providers and to grow. So, sitting down with 19 million users—it was interesting—Allstate started doing this in 2010 and by 2018, maybe we had a million, million two, something like that. People connected, and then with these other relationships now we have nearly 20 million and are processing just an amazing amount of data, which just makes everything better.
Matt Weinberger:
That’s right, that’s right. On the subject of that data, we have a question from Twitter. What’s a surprising fact about driving, or a surprising data point you’ve discovered from sifting all that data?
Gary Hallgren, Arity:
The fun thing is someone said recently that did you ever notice that when you’re driving everyone that’s slower than you is an idiot, everyone that’s faster than you is aggressive, right? It’s like everybody just thinks that their mode is the best. So, people that think they are actually risky drivers maybe aren’t and those that aren’t, or don’t think they are, really are. And so, I think it’s by tying it actually to real claims data and real data, it’s now getting more real and less about just the notion that I think it’s 70% of everybody thinks they’re a great driver. It just simply can’t be true.
Matt Weinberger:
That’s right, that’s right. And then, on the subject of that data, again, right now kind of the hot conversation in Silicon Valley is about regulation, data sovereignty. GDPR is kind of the big one, but all kinds of other potential rules and regulations going to effect. How are you looking at regulations around data as helping or harmful to the business?
Gary Hallgren, Arity:
Yeah, we’re GDPR certified. We’ll be CCPA certified by January. The interesting thing about it, when you step back on it is, like, Do we really need regulation to say, well, if you’re going to collect data from people, you should tell them. If you’re going to sell it, you should let them know. If they tell you to stop, you probably should stop. Right? When you think about it, it’s just data hygiene. So, as we’ve looked through the regulations, and look at how you’re running your business, if you’re not running your business that way, it’s like, you should be. So, we really support these initiatives. We give empowerment to people to control it, and we think it’s all a good thing, and it’s something that I wish the regulation didn’t have to happen to us to do it that way because it’s just the right thing to do.
Matt Weinberger:
That’s right, that’s right. And another question from Twitter. We’ve talked a lot about autonomous vehicles and personal insurance. Do you see application for what you guys are doing around logistics and trucking, and other industries, related industries?
Gary Hallgren, Arity:
Yeah. I spent a good chunk of my career doing the commercial side and right now we’re focused more on the personal line side just because it’s a much bigger market. A lot of the heavy trucks, the Waymo going toward the long haul trucking is, I think, is a fantastic way to start with the autonomous vehicles. To me that makes sense. We do think there’s a commercial application to what we’re doing, but there’s a bigger problem to solve on the personal line side. So, we tend to focus more there. But we’ll get to commercial too.
Matt Weinberger:
That’s right, that’s right. And something else which I wanted to talk to you about is Arity is based in Chicago, whereas a lot of the movement around AI, in general, and smart mobility is happening here in Silicon Valley. How do you think about even just hiring out in Chicago versus the home of Tesla, Facebook, and Google?
Gary Hallgren, Arity:
Right. Well, we do have, corporation does have offices here as well and actually not too far away. So, we do feel like we’re tapping into the talent a little bit here. In Chicago, I think there’s a really nice technology hub, but I tend not to get too focused on any one different place. Across—our management team—we have people from all over the country, and I think you need those diverse perspectives as you’re building any company. I think if you get too insulated out here, you develop a little bit of the Silicon Valley view of things. And I think if we never got out of Chicago, that’d be a problem too.
Matt Weinberger:
That’s right, that’s right. We have several questions coming in about the autonomous industry more broadly. Do you see the autonomous vehicle boom as having an impact on the insurance industry, and how will the insurance industry have to adapt to the rise of Waymo, and Zoox, and several other fine autonomous vehicle companies?
Gary Hallgren, Arity:
Right. I think it’s going to be an evolution. Background—I was a civil engineer, spent time doing traffic engineering earlier in my career, and just looking at what it takes to actually roll things out in cities and counties—that part is very complicated. So, when I look at the journey of what it will take, will there be these vehicles in small, isolated areas? They will. But the demise that people were saying in the beginning that, well, there won’t be an insurance company. So, two things. One is we all do software, a lot of us do software. It’s like it’s going to have bugs, things are going to happen. So, you still need someone to ensure the risk, do you risk it more as an individual versus something broader in a fleet? That could happen. It’s happening today with all the companies that are providing vehicles to the rideshare companies.
Today, they’re doing more of a commercial-fleet type of an insurance solution versus otherwise. But, when I think a lot of what was described in the two videos of having somebody pick me up and bring me to work, and then someone—that person then goes and picks up my kids and brings them to school, and then maybe to my mom to the doctor. It’s like, we have that today. It’s called Lyft. So, it isn’t as though the—I think people are getting very used to doing these new forms of transportation. Whether or not there’s a driver in there, not a driver in there, I think that’s going to be a little bit less of the leap, but I don’t think it’s going to change the insurance landscape that much that fast.
Matt Weinberger:
That’s interesting. So, what is going to change the insurance landscape that fast? What’s kind of an underrated trend that we’re not paying attention to?
Gary Hallgren, Arity:
In my mind, it’s the idea that this, in 2000 and, right around 2000, companies started to realize that if they blended data around credit, together with data that was claims, they had new models and it created insurance companies, and those that won, won big and those that didn’t, lost big. And that’s what’s happening again with these driving scores. It used to be that we had an idea that driving was going to be more predictive than anything else, and now we know, not think maybe might be, but actuarily know that it is. So, the fact that we have the scale, we think that every insurance company’s going to start to rate this way, and it’s going to really have great effects I think for the ridesharing groups, and all the new forms of transportation.
Matt Weinberger:
That’s interesting. And then, kind of along those lines, what do you think is the most misunderstood thing about what you’re doing, or AI and data collection, in general, that you wish was more understood by the world at large?
Gary Hallgren, Arity:
The—I mean, the big-brother side of things, the idea that this is some sort of—that people aren’t going to respond well to it, and I think that that’s really going to change. I think we need to do it that way. I mentioned that I want my Lyft driver to be a good driver. It’s like you want that. If I’m going to be driving your car some time, you’re going to be sharing your car out to somebody else and leaving it at the airport—everyone’s going to want to know these things, and I bet if we look five years out in the future and look back, it’s like, of course, you’re using driving data to do this. To me, it’s more empowerment, it’s more coaching. There’s ways to improve. It’s more transparent. So, everything about it, I think, is a little bit misunderstood today, that if we go out in the future, people just accept it and embrace it.
Matt Weinberger:
Awesome. That’s right. That’s right. Awesome. Well, I think that’s about it for our time. So, thank you so much for being here today.
Gary Hallgren, Arity:
Awesome. Thanks, man.
Matt Weinberger:
Thank you.