AI Wisdom Ep. 15: AI in the Driver’s Seat with Roberto Sicconi

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.

Today, in the United States, trucks with human drivers behind the wheel move 70% of the nation’s goods. But this is set to change with a wave of new technologies like AI, computer vision, machine learning, and autonomous driving, all pulling together behind the wheel. And let’s not even forget, Uber, ridesharing, Lyft, and all of these amazing technologies, all of which are going to have a major disruption in the coming years. I’m very excited to be here with Roberto Sicconi, CTO and Co-Founder of dreyev, who’ll be joining us to talk a little bit about what insurance means in the age of autonomous vehicles. Roberto, thank you so much for joining us. Can you introduce yourself and tell us a little bit about dreyev?

Roberto: My name is Roberto Sicconi. I’m CTO and Co-Founder of dreyev. The company has been operating since 2016 and is specializing in solutions to prevent distracted and drowsy driving. We are offering at this time a solution as an aftermarket device that can be installed in vehicles, that can be cars or trucks. In the future, we plan to license technology to car manufacturers, so our solution could be used on any modern car.

Ron: Great stuff and I can only imagine how many lives it’s going to be able to save. So, I’d love to jump right in and talk about the Uberization of trucking. How do you see this trend transforming the almost 1 trillion, roughly $800 billion in the U.S., industry, and what is the impact that it’s going to have on commercial insurance for trucking?

Roberto: Well, commercial transportation is a pillar of U.S. industry. Pretty much every sector depends on it. One of the things that is happening is that more last-mile delivery, more local delivery, has taken place in the last few years, particularly this year. Also, thanks to the spread of the coronavirus, and that is forcing companies to rethink the way they deliver the goods in some cases straight to the home of the customers who are ordering them online.

Think about the impact of online commerce, particularly with Amazon, Walmart, and others. This has pushed up the needle substantially. The number of drivers delivering smaller, medium-sized trucks in urban environments, has increased substantially, changing the nature of transportation. That also has implications on how you select and train drivers.

There is also an overall change in the transportation industry where long-time experienced drivers are retiring. The electronic logging devices’ main data has sort of discouraged, to some extent, the longer driving that the older drivers used to use to improve their salaries. There are many more young drivers coming into the workforce, but many of them lack the experience, and as a result there is an increased number of crashes, particularly caused by distraction.

Ron: That’s very unfortunate to hear. Now, I’m curious because I’ve heard a statistic, and you probably have more up to date numbers: I heard at one point that something like 40% of the U.S. is employed in an industry somehow related to trucking, whether that be logistics, warehousing, or of course truck drivers themselves and everybody else who has to support them, crane operators, etc. When you think about AI and self-driving cars, and I think, you know, cranes could potentially be automated as well, although I’m not an expert on the topic, what do you see as the change? Should people be afraid? Or where do you see the industry going?

Roberto: It depends on the horizon. I would say for the next 5 or 10 years, I don’t see a major change in the way trucks are being driven around, especially local delivery. Long-haul trucks may see some initial attempts to bring in driverless vehicles, driverless for a portion of the trip that might be on long stretches of highways. But when you have a lot of traffic or you’re inside dense communities, I don’t think automated, especially large vehicles, have any possibility to take over traditional ones. There is definitely a trend in bringing in automation wherever possible. You mentioned cranes. Cranes, maybe not so much. But at harbors, for instance, there are autonomous vehicles that transport containers automatically. And that’s usually in the areas where there are not many people around, and the area is free of regular traffic. It’s on dedicated vehicles that move around and they can take each other into account.

I think that AI is probably going to come in mostly in terms of help and a sister to human drivers to make sure that they minimize their mistakes. And it can also help reduce fatigue if some sections of the driving can be handled by the vehicle. But still, I think we are still some time off.

Ron: If you were to venture a guess – and I recognize that this is very much an exercise of looking into a crystal ball and predicting the future – when do you think autonomous vehicles will become ubiquitous?

Roberto: If I look at what the leaders in the space are proposing, I would say, 5 to 10 years is when you will start seeing some driving by select vehicles on select arteries, and they will probably cover only a small percentage of the overall traffic. So, a huge, pervasive replacement of vehicles capable of driving by themselves. I don’t think it’s going to happen before at least a decade and a half from now. And if you think of shorter segments, I think it will take even longer. The same applies for personal cars.

Ron: I mean, that’s interesting. Two threads that I want to pull on there. I’ll start with them in order. So, something that you said a little earlier on, you mentioned that today there might be some automation already on major highways where there aren’t necessarily on-ramps and off-ramps. And you mentioned specifically that you wouldn’t trust a big rig to operate in the city. And I thought there was this distinction there between highway operation and inter-city operation. Do you think the public is ready for self-driving vehicles on the highway at least?

Roberto: Well, according to polls, the majority of the regular public is very concerned. So, there are certain people who are more ready than others, but I think there’s still a minority, probably 20% or 30% in terms of automation for their own vehicles.

In terms of accepting driving on a highway where there’s a lot of driverless vehicles, I think the percentage is probably even lower. You have to see this work well before you can trust it.

Ron: Is there, in terms of difficulty, a distinction between the inter-city versus highway driving?

Roberto: Well, traffic patterns are very different. Of course, on a highway, a crash can be a lot more costly both in terms of damages and people lives. But they’re much less frequent.

So, the reason why automation is pushed first on highways is that in normal conditions, even if you’re driving at a relatively high speed, the frequency of crashes on the highway is lower and the traffic is a little more in control. In a city you can have vehicles moving around unpredictably, coming from all directions, and you’re much more exposed to crashes, maybe less damaging because the speed involved is lower, but much more frequent. And the ability to deal with traffic in the city I think, goes way beyond what is required on the highway.

Ron: The other thread that I’d love to pull on is to talk a little bit about Tesla because, I think, at least a lot of people over the recent years have become more comfortable with Teslas. They’ve seen Teslas and obviously we hear about the unfortunate Tesla crashes. But I think when you look at the National Transportation Safety numbers, supposedly it’s the safest car ever. It was like the only car that ever scored a five, I believe I saw once. And something that a lot of people would say is very important, and on top of that, you know, it has the self-driving capabilities. Obviously, it’s not Level 4 autonomous, what they call it, I believe it’s only at Level 2 autonomous today. But can you speak a little bit to that?

Roberto: Yes. And as an engineer, it’s hard. I admire what Tesla is doing. It’s a very bold approach to new generations of vehicles. I think what they’re doing is, in a sense, both a blessing and a curse. The Level 4 autonomous driving is not ready, anything short of it can create potential issues, and potentially big ones. And the problem is not so much because of the limited capability of the vehicle, but because of the misunderstanding of the driver about what the capabilities of the vehicle are.

So, the underlying assumption, even in today’s Tesla, is that the driver is always paying attention. And unfortunately, that was not the case. I’ve seen plenty of videos on YouTube with people showcasing how well the car is doing whilst doing the most stupid things. If something happens that the car could not anticipate, and they’re not ready to take over, bad things can happen. And the more you rely on your car being able to drive you all the time, the more you expose yourself to these kinds of crashes that are highly preventable.

So, I think the biggest limitation right now is in the fact that the system that Tesla is putting in cars is not really monitoring the level of attention of the driver. They expect the driver to put their hands on the steering wheel, every now and then. It can be encouraged. The mechanism that they’re using today is definitely not foolproof and you have seen many cases where people were, you know, putting some weight on the steering wheel to simulate the hands on it and then get into a crash.

There has to be something that overcomes, if you’ll allow me the word, the stupidity of drivers who are over-relying on the capabilities of the car to make sure that those things don’t happen.

Ron: I 100% agree. And I’d love to talk a little bit about dreyev, because from my understanding, it’s an intelligent crash avoidance solution for fleets, an insurance that basically puts an AI-powered co-driver in the passenger seat to help manage the behavior and evaluate the attention as you just talked about. And really, that means fewer crashes, which means fewer claims, which means better loss ratios. And so, what kind of interest are you seeing both from the carriers as well as the fleet companies themselves? And then finally, I’d love to hear a little bit about if the ridesharing companies are interested in this as well?

Roberto: Yes, attention monitoring is a hot topic. It has been around for a long time. I think there’s enough consensus now especially in the commercial vehicle space that the only way to really understand what the driver is doing is by observing the driver.

We can extract much information out of telematics-related information about braking, cornering, speeding, or so on, but we don’t really know if the driver is truly paying attention to the road or not.

He may have done the right maneuvers by luck, or he may have done the wrong maneuvers, but he was still in control and so the level of risk was low.

In order to understand how much risk the driver was taking at a given time, you really have to observe the driver. And the best way to observe is by using a camera that looks at the face of the driver, maybe even the hands, and then puts that into context.

It is very different when you are driving in Nebraska on a straight highway when there’s no traffic and you can look around and enjoy the landscape than it is doing the same in a busy environment where somebody might be stopping before you, you don’t see it, and you crash into it.

So, having a notion of where the other vehicles are, how far they are, in what position they are – are they coming into the lane or moving away? – and whether you as a driver are observing them or not makes a complete difference. And this is what we want with dreyev. We want to make sure that we understand what the driver is doing at any given time, in the context, and then based on that, we warn the driver when there’s a need for it.

Ron: Amazing stuff. And so, what type of interest are you seeing on that?

Roberto: I think there’s no doubt that fleet managers are looking more and more into video-based solutions. I think the first step that has been pioneered a few years ago, and which is starting to become a little more popular, is to record what the driver is doing so that, in the unfortunate case of a crash, you can go back and see what the driver did, whether it was the right maneuver or not, whether he needs to be coached, and who gets to be blamed. In some cases, the videos allow you to exonerate the drivers.

But it’s postmortem, if you will allow me the word, analysis. I think the real big and new step is the ability to detect all of this in real-time, actually before real-time in a sense.

So, when you see that the level risk is growing and the driver is not paying attention, you see, I have a little chance to warn the driver and have the driver take a maneuver and then avoid the crash. And that, in our mind, makes a big difference from knowing exactly what happened after things have already happened. So fleet managers are looking into this. Some of them that we are speaking with are willing to go with our solution over a video-only-based solution they had before. And we expect this trend to grow. Insurance companies, especially commercial insurance companies are looking at this. Personal insurance companies are looking at this too but are worried about the cost of the equipment. So, we are looking at solutions to cut down the cost and make it possible in regular cars. But the needs and the value of the camera, I think, are out of question.

Ron: So that’s interesting. Does it require additional hardware? And just to be clear, is that additional hardware already present in commercial trucks, and it’s simply not present in personal vehicles? Or, it’s not present in commercial trucks, but due to the economics it makes sense to, you know, postmortem, to use that word, install them aftermarket?

Roberto: Yeah, generally speaking, no vehicle comes with a camera embedded. Many vehicles, particularly personal cars, have camera-based solutions to look at the road and those are used for driving assist. So, for lane departure warning, for collision warning, automated braking. Those have been around for a few years and there’s plenty of cars where you can buy it and they help you, but none of them... Actually, I shouldn’t say that. Some of them have a camera that looks inside, but very few. Subaru has one model; the Forester carries it and now they’re expanding it to the rest of the line. Cadillac used to have a model. There are a few high-end cars that have this. In terms of commercial vehicles, as far as I know, none of them do. So, in all cases that I’m aware of, commercial vehicles can be retrofitted with the cameras looking at the driver. But it’s additional hardware that needs to be installed in the vehicle. Not difficult to install but it doesn’t come with the vehicle.

Ron: I’d love to change topics a little bit and get your thoughts on ridesharing, which is another part, obviously, of commercial transportation, although very different. And the risk there, especially the share in the liability is very different than trucking. You know, what is your experience? We’d love to hear any thoughts that you have on how that liability is shared, how insurance companies feel about it, and how AI and machine learning can play a role.

Roberto: The first thing that comes to mind when talking about insurance for rideshare drivers is there are select insurance companies that offer policies for rideshare drivers.

I know that there are a lot of drivers who are using their own car with their own regular insurance without declaring to the insurance company that they’re using it for occasional ridesharing. And when they do that, they put themselves at serious risk because, if they get into a crash, and then the crash happens to involve some other passenger that was not supposed to be part of a family or so on, they may not get the coverage.

So, one of the things that I believe Uber, Lyft, and other rideshare companies do is offer special additional policies for drivers that cover them during the rides. And then you have to demonstrate that you were on a ride with somebody in the vehicle rather than approaching the place where you wanted to pick up the customer, but you were on your own.

So, there are some tricky situations that I don’t think are always sorted out properly. There are packages, particularly in New York, I believe, where a rideshare driver, who does it as a profession gets covered all the time. In most cases, it’s a little bit of a patchwork. So, you have your own personal insurance, which may not cover you, if they discover that you were doing a rideshare drive and something happened.

Ron: So, one question that I know a lot of people would love to know and I’ve heard very, very different takes. And so I’ll preface it by saying “to be determined,” and I’m not looking for the definitive answer here, but I’d love to hear in your experience, who pays for insurance and who’s liable? Especially when it comes to self-driving technology. So, you know, you might have a Toyota manufactured car. And I know Toyota in this case might be working on their own technology but let’s just say in this world, they license the technology from Tesla. Who pays the insurance premium? Is it bundled into the car? Is it separate? Is it a separate insurance company? And then when a claim gets filed, you know, is it against Tesla? Is it against Toyota? Is it against the insurance company? Any thoughts on that?

Roberto: I do not own a Tesla, so I don’t know firsthand, but I have heard from others who do.

Typically, if you want to insure, with a regular insurance company, a Tesla car that has self-driving capabilities, your premium goes up significantly. And Tesla is now offering their own insurance for that reason, because otherwise some of the customers may walk away from purchasing a car.

Going forward, when there’s going to be multiple manufacturers, not just Tesla, I think it will depend on the manufacturer. Remember years ago, Volvo and Honda both announced at some point when they would feel that the car is safe enough to drive by itself, they would take over the insurance portion of the driving done by the car.

So, in that case, you would have to essentially cover with your own personal insurance only when you drive but when the car drives itself, any liability goes to the manufacturer.

In principle, the problem that I see coming is what happens in the transition. When you reclaim control of the car, the car was driving by itself, but you feel like driving yourself, so you take over control. What takes notice of that so that if something happens because you took over and you did the stupid thing, you got into a crash, the car manufacturer does not get liable and vice versa? And how do you know that the car is not relinquishing control to you because it feels like it cannot handle the condition and then you are left with doing what you can with the car, and then you are liable?

I think this is going to be a big debate in terms of who is responsible for what, and why, and how did the change in control happen? Who requested it? Who pushed it? Was there a consensus? You know, when you have a fraction of a second to make those decisions, thinking of your insurance and liabilities is the last thing that comes to your mind.

Ron: Very well-said. I love that. So, we’re going to take a quick 20-second break to tell you where you can find 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.]

Ron: We’re back with our featured guest, Roberto Sicconi. Let’s jump right into the next question. I don’t think there’s any doubt that technology disruption continues to be a dominant theme for the insurance industry. When you think about insurance companies that are struggling with change and embracing new technologies and business models, what are your recommendations for them?

Roberto: I think we are at the verge of a major change, probably disrupting capability or opportunity for insurance companies to jump in. And that’s not just for commercial vehicles, or commercial insurance, but also for personal insurance.

Historically, the cost of a policy, the premium, was determined based on statistics that had little to do with your driving skills, really. Most of them were related to where you live, where you operate, the size of the car, the speed, the color, whether you’re married or not, how long you have been driving, whether you had crashes in the past. All of them are meaningful parameters, but they don’t really tell the whole story.

And you could be an extraordinary driver most of the time and then occasionally you slack off and do stupid things. And the insurance is not reflecting it. And you get no notion of the fact that you’re taking more risk, and therefore, as something happens, there’s a lot more liability associated with it.

So, I think that in the process of moving to everything online, everything real time, the idea of having policies that can change over time, depending on the type of risk you take, and that could change every month, every week, every minute, potentially, as long as you know what you’re doing and what the cost is going to be to you, is not handled properly by traditional insurance companies where you renew the policy every year.

I think that usage-based insurance is making a first step in the direction. It does not really translate risks that you’re taking to price changes. That mechanism can be refined by observing how you are doing, how much risk you’re really taking. But as long as there is a feedback loop that you get informed and insurance gets informed, I think we can go towards a much more dynamic pricing scheme where prices are calculated based on risk that people are taking every minute they drive. It’s very different from the situation we have today. And I think that insurance companies who are willing to embrace this new concept are going to be way ahead of anybody else. And they will automatically get access to the best drivers, and the lousy drivers who are really the most expensive drivers that now they have difficulty locating will end up with insurance companies that will have no choice but to close because they cannot afford them.

Ron: So, it sounds like it’s a competitive advantage. And from your perspective, to be able to have this usage-based pricing both for the consumer, because as you said if you are safe, you should experience reductions in premium ideally. And on the flip side, it’s also beneficial to the company because if they are following this, they’ll be able to attract the best customers, reduce loss ratio, and obviously increase their book of business.

Roberto: Correct. And I think that the flip side of this new coin is the fact that if you get dynamic information from the insurance about the risks you’re taking, you get guidance on how to change your habits, so that you minimize the risk unless you want to take it. But let’s assume you’re against it, you may not even realize how much risk you’re taking. The insurance can measure it and let you know, then you can change your habits and then eventually you’re going to be a better driver. And as a result, everybody gains.

Ron: That would be amazing. I’d love to take a pivot here to talk a little bit about your startup journey. You’re a fellow founder, and I’d love your thoughts on what the current situation means for startups, especially as we all deal with, you know, the ongoing impacts of COVID-19.

Roberto: Well, that’s a very interesting topic. It really depends on what your startup does. If you’re on online business, particularly if you’re into, I don’t know, video conferencing, then your business is booming. If you are in traditional businesses, it may be booming if you are into online booking delivery, and it will be tanking if you’re into anything that requires a brick and mortar shop or manufacturing other things. In our case, we are in a kind of in-between condition where the market is under pressure and there’s no doubt that transportation is extremely important, especially for the goods in high demand.

Even grocery deliveries are up because people are stuck at home. So those kinds of fleets are under pressure to deliver more and we get requests to handle that.

Other ones are delivering, for instance, cars that were not produced because the manufacturing sites were closed. They have nothing to do and they have been laying off people. So different industries have different problems. In our case, we see a striking difference between companies that we work with that have nothing to do or not much to do with the virus’ spread, they can’t continue to work. And others were affected very badly like the motor coach business – whenever you have passengers that need to sit closely inside the same cabin, they cannot operate now. So there, it definitely has come in as a tsunami, and we are dealing with this.

Speaking for all startups, some of them can benefit, some others will suffer. But it’s a very challenging time for everybody.

Ron: So, there’s a lot of debate and we touched on this a little bit earlier about how AI is going to alter the future of work and some fear around AI replacing human workers. Your solution is really focused on augmentation rather than replacement. It’s about putting a human co-pilot in the vehicle. How do you see AI augmentation affecting the workforce?

Roberto: Through augmentation, I think it can be beneficial in every industry.

There’s no doubt that intelligent machines, however you want to picture them, have helped productivity in all sectors.

And yes, some of the jobs have been replaced but usually, there’s a net positive by creating new jobs where people have to learn how to deal with machines and machines do the hard work. I think this is not going to be the same here. There’s a lot of advantages of using machines at your disposal so that you can be more effective as long as people get used to it and learn how to do it and machines are designed to be helpers rather than replacements.

I think this is a fundamental question to anybody who is designing AI-based solutions going forward.

One of the issues that I see coming is that, as AI becomes virtually as capable as human intelligence, we need to make sure that the ethics around it or behind it are solid. So back to the self-driving car, you want to make sure that the car is behaving the way you would. If it’s given an unexpected situation, it’s not going to kill people simply because it was designed that way.

So, there are, I think, fine boundaries that we have to analyze very carefully and making sure that any decision-making that is built into a AI-powered machine is scrutinized and understood. The last thing you want to do is create black boxes that get trained, and you supposedly think can handle all the conditions the way you want, and then all of a sudden, they go out of control and you don’t even know why. So that’s what is called transparent AI is a movement inside the engineering community and I think there are very strong guidelines that need to be adhered to. The UK is pushing very hard in this direction, I think it’s a very important step to make.

Ron: And for business people, it’s very similar to a cognitive bias or a data bias where you take a sample, sometimes, and it’s not the only way, I think you’re referring to explainability. But there’s many ways that an AI can be trained in a way that’s not ethical or might not work well. And so, I love that you guys are conscious of that and working to make sure you guys aren’t doing that.

So, as we wrap up, I’d love for you to share one piece of wisdom for our listeners, and it doesn’t have to be business-related. It can be life or in general, you know. What’s on the top of your mind that you’d like to leave everybody with?

Roberto: I would say be positive. I mean, this particular period we are going through is difficult, it’s challenging, it highlights issues that have been lingering for a long time that nobody ever analyzed, and all of a sudden, they have come up like in healthcare, for instance. But that’s a good time to rethink strategies, to rethink technology, rethink the way we deal with each other, rethink what we think is valuable in our lives.

One of the things I learned was, for years I postponed getting in touch with friends and family because there was always something else that was urgent, and then all of a sudden, I realized that’s actually the most important thing in my life. And having a little more time or being forced to use more communications with people whilst waiting to figure out how to deal with the expansion of the virus, tee up some of the ideas and I think have a much stronger priority list in mind right now. And going forward I think society should do the same as a whole.

Ron: I love that. Stay positive. And I’m very glad to hear that, you know, this time, even with all of its negatives, has brought some positives forward for you. So, Roberto, thank you so much. If people want to find out more about you and dreyev, where can they find you and connect with you?

Roberto: Well, the easiest way is to go to our website, dreyev.com. We have links inside. There’s a way to ask for a demo. Get in touch with us, and we’ll be happy to give you additional information and documents.

Ron: Amazing. Thank you, Roberto. And as always, you can find us at www.chisel.ai, on LinkedIn, make sure you like and subscribe and we’ll have another episode for you shortly. Thanks again, Roberto.

Roberto: Thank you so much for having me.

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: SoundCloud, Spotify, iTunes, Google Podcast, 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!