Driving You Mad
One mid-October night in 2015, owners of the Tesla Model S received an over-the-air software update. It wasn’t so much the way the software was delivered that made the update news.
The electric cars regularly receive the latest features and functionality this way, preventing a trip to the dealer.
What was remarkable was what the software release, which the company considered a public beta, allowed the owners of the cars to do. The update enabled a fully integrated autopilot system, with the cars receiving feedback from radar, ultrasonic sensors, forward-facing cameras and GPS. The autopilot system enabled the car to steer within a lane, change lanes on its own when the driver activates the turn signal, and adjust speed while in cruise control. Not only that, it could scan for a parking space and parallel park. With the driver’s permission, data from the car is anonymously sent back to Tesla, in effect turning the owner into a software test driver, helping further hone the car’s systems.
The hardware that allows the technology to work has been incrementally introduced into the vehicles over the past year, according to the company, and other upgrades have happened since. The driver, Tesla says, is “still responsible for, and ultimately in control of, the car.”
But with such updates now available literally overnight, how long will that be the case? Tesla anticipates a fully automated car within the next three years.
Autonomous or “self-driving” vehicles used to be the stuff of the future. But they’re just one way the automotive industry is turning the “far off” into the “right now”—and affecting numerous other fields along the way.
How long ago could our fiercely independent American culture have imagined the growth of a car-sharing company such as Zipcar or a ride-booking application such as Uber? When did we first imagine we could summon someone else’s vehicle via an app, that a generation like the millennials would be driving less and buying fewer vehicles than their predecessors, or that cars would become so advanced that liability would begin to shift away from the driver in the event of a crash?
For drivers, and for the auto industry, it’s an exciting, rapidly changing time. But at this juncture we must be cautious, experts say, not to take our eyes off the road or our hands off the wheel. Instead, we should pay full attention to the direction the industry is headed.
“Together with cognitive computing and artificial intelligence, this will be the next industrial revolution,” predicts Sebastiaan Bongers, the manager of automotive strategy at Swiss Re. And the impact on society, Bongers says, will be immense.
Naturally, not all of the talk about autonomous vehicles is personal in nature. There’s the question, for example, of whether our freeways will soon be stocked with round-the-clock driverless commercial trucks. Bongers, for one, doesn’t anticipate they’ll be adopted by all countries for at least the next couple of decades.
“It is more likely that we will adopt an airplane-like model in which a truck driver will still be required to oversee the truck’s actions,” he says. “And truck drivers cannot stay awake 24 hours. There will be limitations. But if regulations were to change, the sky is the limit.”
Change already is happening. Last year brought news that driverless trucks had been operating for several years at the iron ore mines of Pilbara, in Western Australia. Could driverless trucks be the solution to the shortage of truck drivers in the United States? In Pilbara, the goal was to increase safety and efficiency—though the roads the trucks travel are relatively deserted.
Now, from one extreme to the other: last fall Daimler began testing semi-autonomous trucks on Germany’s Autobahn, guided by sensors and radar.
Audit, tax and advisory firm KPMG, which released the report “Automobile Insurance in the Era of Autonomous Vehicles” in 2015, speaks of recent changes causing a “marketplace of uncertainty.” For one thing, as autonomous vehicles increase in number, crashes will be dramatically reduced. Accident damage payouts and premiums will follow.
“Premium follows loss costs,” the report reads. “Whether personal or commercial, auto insurance is a mature and competitive industry, and given these market potential dynamics, it would be naive to think that premium will stay the same while losses drop, thereby dramatically increasing underwriting profit for carriers. To the contrary, KPMG believes that carriers could drop price in order to stay competitive. Plus, consumers will no doubt demand premiums to reflect fewer accidents.”
The report continues: “As the size of the market shrinks, we anticipate the potential for frenzied competition as firms attempt to maintain premium volume to cover operational expenses or market share. Carriers could potentially lose sight of pricing business for profit. This irrational pricing behavior by either well-capitalized or troubled countries could result in a dangerous downward underwriting spiral for the broader industry. Things could get ugly.”
Getting Ahead of the Curve
Donald Light, director in Celent’s North American property-casualty practice, authored a 2012 report called “A Scenario: The End of Auto Insurance.” The report considered the already present use of telematics, collision avoidance and automated traffic law enforcement, along with the emerging technology of “robot cars,” on the insurance industry. Light’s scenario foresaw wide deployment of those four technologies, radically reducing the frequency and severity of vehicle crashes and, as a result, substantially reducing the level of auto insurance premiums.
When the report came out, Light says, most people rejected it as unlikely. Some said increased technology would make cars more expensive to repair, so even if the number of accidents went down, the revenue base would remain the same—or actually improve. The impact, then, would be negligible.
In the past couple of years, however, that perspective has changed. Cars have become increasingly connected through telematics, with much more data available for insurers to set premiums at levels which can incent better driver behavior and, ideally, lower loss experience. The data can also provide a feedback loop to help manufacturers design better products. Advanced safety features, such as sensors and other technologies that have been proven to reduce crashes, already are becoming standard on an increasing number of models.
In the midst of these changes, the burgeoning shared economy has continued to grow. Companies such as Uber, Lyft and Zipcar are increasingly making a case against car ownership.
“From an insurance point of view, that means fewer cars on the road,” Light says. In large, densely populated cities, he sees a feasible reduction of 20% in the near future, which would throw the insurance industry off kilter.
Add it all up, Light maintains, and the future of auto insurance is “dire.” Like everyone else, he’s unsure exactly when the anticipated substantial drop in insurance premiums will happen—or if it’s soon enough to provoke a call to action for current agents and brokers.
A lot of companies, he says, are now thinking about getting in front of the curve.
“Three years ago,” he recalls, “they were saying, ‘There is no curve.’”
We find ourselves, then, in a transition period, in which auto manufacturers are looking ahead, trying to assess demand; in which consumers are growing more dependent on whiz-bang features; in which drivers of traditional cars and ever-more automated vehicles will need to learn to coexist with each other; in which technology outpaces legislation; and in which the impact on future government revenue from decreased DUIs, speeding tickets and other traffic offenses is unknown.
Light likens it to the fog of battle, where in the midst of the fight, it’s difficult to predict when the war will be over, even as it starts to become more certain who the eventual winners and losers will be.
Liabilities and Legislation
It’s hard to enter any conversation about partially or fully autonomous vehicles without mentioning safety. Significant reductions in crashes and deaths are a strong driver. RAND Corporation’s 2014 report “Autonomous Vehicle Technology: A Guide for Policymakers” notes there were more than 5.3 million automobile crashes in the U.S. in 2011, resulting in more than 2.2 million injuries and 32,000 fatalities. If all vehicles had forward collision and lane departure warning systems, blind spot assist and adaptive headlights, according to the Insurance Institute for Highway Safety, nearly a third of those crashes and fatalities could be prevented. Fully autonomous vehicles, according to the report, would further affect statistics, due to the correlation of driver error and crashes.
For that reason and others, James Anderson, a senior behavioral scientist and one of the report’s authors, says he senses “a lot of curiosity and excitement” about increased automation and self-driving cars and not much fear.
“It wasn’t that long ago that people were concerned about elevators operating automatically without a human controller,” Anderson says. “The research so far has shown that people quickly get over it as they become accustomed to how these things work.”
Here, Anderson sees two separate approaches. First, there’s the incremental mode, in which the gradual introduction of features such as adaptive cruise control and lane departure warnings help warm drivers to the idea of releasing control.
“This strategy makes a fair amount of sense,” he says. “And it allows for automated driving in limited circumstances, with the driver ready to take over. But research suggests that humans over-rely on these things pretty quickly. The real challenge is keeping the driver engaged.”
And that reality provides support for the other approach, taken by Google. The technology company’s self-driving car project, founded in 2009, had logged more than 1.2 million autonomous miles by October 2015, with 10,000 to 15,000 more miles being added each week on public streets. The idea is to “sidestep all of the human and computer interaction issues and jump right to a model that doesn’t have a steering wheel at all,” Anderson says. “You design a system that’s robust enough that it doesn’t require a human to intervene in all but the most extreme circumstances. This actually addresses some of the safety issues by not having human backup, but the car has to be in a somewhat protected area or driving at a relatively low speed. I think we’ll see both of these models progress and probably merge at some point.”
And speaking of merging, there is much talk, Anderson says, about the way autonomous vehicles and ride-booking services will combine in the future.
“It’s not a coincidence that Uber recently hired a good chunk of the Carnegie Mellon University robotics department to work on this stuff,” he says. “They’re thinking the same thing. There’s a lot of potential there—the ability to completely shift the way we think about transportation.” Instead of automobiles being privately owned, transportation-as-a-service could become the new order of the day, when consumers would simply summon whatever vehicle they desired, wherever they desired it—“and all the hassles would be someone else’s.”
That’s right up the alley of Susan Shaheen, co-director of the Transportation Sustainability Research Center at the University of California, Berkeley, who focused her doctoral dissertation on “smart” car sharing, linked to public transportation, in the mid-1990s.
“In the early years, car sharing emphasized testing models, demonstrating program viability, and mainstreaming sharing,” Shaheen writes in an email. “Over the years, we have seen the conversation shift from car sharing to the broader sharing economy. Sharing has become more mainstream. Users are looking for access across a wide spectrum of services, including round-trip car sharing, one-way car sharing, ride sharing, e-hail, and ride sourcing (e.g., Uber, Lyft). Car sharing has played a key role in mainstreaming shared mobility more broadly.”
Even so, misconceptions remain.
“A very common misconception is that car sharing is only viable in mixed-use urban settings,” Shaheen says. “In the early years, this was the most common deployment of car sharing. However, over the years new service models have emerged, providing shared vehicle access in a variety of contexts, including fleet sharing for public entities/businesses and college/university car sharing.”
So does she foresee a day when no one actually owns a car, when all transportation is shared? That’s still a bit unclear. Getting people to give up their vehicles may take some convincing—especially if they don’t fully track or understand their transportation costs and miss the fact that pay-per-use might well be less expensive. All the same, she says, automated vehicles “can definitely enable new opportunities for vehicle sharing and reduced vehicle ownership.”
“Taxi, shuttle and carpooling services may be able to automatically pick up and drop off riders along digitally dispatched routes, increasing vehicle occupancy,” she says. “It also may facilitate prearranged rides (e.g., a vehicle syncs to a person’s schedule and is dispatched when it shows an appointment in someone’s calendar). This is important because there may be opportunities to link calendars together and dispatch shared rides (e.g., two people at a large office complex are going to meetings in the same location).” In addition, within individual households, one spouse may go to work, then send the vehicle home to pick up the other. This would reduce parking at the origins and destinations, she says, but may not necessarily result in lower vehicle miles traveled.
Before we reach that point, however, Shaheen sees a need to focus on peer-to-peer car sharing and insurance first.
“While California, Oregon and Washington have state legislation that enables private vehicle owners to place their vehicles into car sharing (similar to Airbnb for renting rooms), most states do not have policies that allow these transactions,” she says. “This could result in gaps in coverage for peer-to-peer car-sharing vehicle owners and users. Developing a model policy framework or national legislation could support the growth of peer-to-peer car sharing.”
Kara Kockelman, a professor of engineering at the University of Texas at Austin, also has been keeping a close eye on this sharing economy and the expanding world of automation. Her 2014 research, along with then-doctoral student Dan Fagnant, envisioned networks of shared autonomous vehicles (SAVs), in which consumers would pay an initial subscription charge, then pay a fee per use. The SAVs would be safer through the use of collision avoidance systems, would reduce labor costs since no paid drivers would be needed, and would reduce emissions thanks to fewer cold-engine starts.
The scenario, however, is not a panacea. Easier travel is likely to mean more vehicle use, longer travel distances and potentially more congestion. Also, Kockelman notes, “We do like storing items in our cars.” Items that pose a particular challenge are the various types of car seats for children. Their availability in SAVs remains to be seen.
She does imagine that some would not want to give up the joy of driving or the status symbol of owning a particular vehicle, “so you may not see people going completely into shared self-driving vehicles. But I do think, once SAV use catches on, it will become pretty popular. As for self-driving vehicles, they may become required in some locations, like busy downtowns with lots of pedestrian activity and high crash histories.” It’s also possible, if self-driving vehicles take over, that there would still be special tracks—akin to amusement parks—on which people could still enjoy the driving experience.
Traditional manufacturers might like authorities to create, for example, special lanes where their cars could travel. But Kockelman says, “Most of the leading-edge car manufacturers say, ‘We know that we can’t expect departments of transportation and cities to design the way we would like them to design. We have to build a car that can handle anything.’”
Naturally, that’s what Tesla—and others—are working toward. But Tesla’s recent upgrade caught many by surprise, Kockelman says, including regulators. It wasn’t just because Tesla already had enough technology on the vehicle to allow drivers to feel comfortable with the automated capabilities. It was because there was enough engineering at headquarters that the update could be done so quickly over the air.
“That’s phenomenal and scary and exciting,” she says. “There isn’t any legal standing on the books for how this all should work and how it should look, and if one of those gets in a crash while self-driving, that could be a problem.” The Model S, she notes, can’t stop for stop signs or red lights on its own at this point or change speed for different speed limit settings, so it can’t really self-drive.
Long-term, Kockelman says, widespread fully autonomous driving would mean the more efficient use of intersections as well as cars that would better communicate with each other.
“We do worry about the congestion impact of cars being driven cautiously because they’re on their own,” she says. That’s one of the issues being seen with the Google car: some drivers can get upset when they’re driving behind it.
“What I think we don’t want to see is a lot of these vehicles driving around empty, unless they are these particular kinds of shared-fleet vehicles, where they’re regulated and they’re allowed to travel maybe 10% of their total miles empty to pick up passengers,” she says. “But I think we don’t want to see people driving downtown and telling their cars to go home or to go four blocks away to park at a cheap garage. I think that behavior can become problematic in all sorts of ways.”
Thinking Differently—Even Now
So what does all this mean for insurance, both now and in the future?
For a start, Swiss Re recently identified several key trends for insurers and reinsurers: the cost of accidents will rise; personal liability will shift to product liability; deep-pocket liability may increase for defendants; car manufacturers will manage and pool risk; and the nature of retail insurance will change.
According to Swiss Re’s report “The Autonomous Car 2015: Mapping the Road Ahead for the Re/insurance Industry,” insurers will need to “think more broadly about their product offerings, including offering pay-per-journey insurance or providing cover to manufacturers for technical defects or carpool owners rather than individual drivers…. Although it is uncertain how legal or regulatory issues will play out, we do know that our role as reinsurers and insurers will change considerably. Many of these changes may forge new opportunities, and those who can adapt by targeting new market segments and diversifying their product offerings may stand to benefit from these opportunities.”
Celent, too, offered its take in mid-2015 with the report “Insurers Beware: Speedbumps on the Road to Autonomous Vehicles.” Here, author Craig Beattie states six reasons insurers should be concerned about the new variables of software and hardware involved in parts of the driving activity. First, there’s technology failure, or expectations about the car’s capability and the driver’s responsibility. Then there are the costs to repair new technology features; the driver’s ability to assess risk; deliberately distracted drivers; and the distractions of driver alerts. Finally, as confidence in technology makes people believe they cannot have an accident, the perceived value of insurance will decrease.
“There certainly are changes in technology, and advances like lane departure warnings and collision avoidance systems are becoming more and more standard,” says Sandee Perfetto, the director of personal auto/umbrella product development at ISO Insurance Solutions, Verisk. “What we’re looking at, and what insurers need to be looking at, is how these advancements in technology and safety systems affect the underwriting and rating of a vehicle and ultimately the claims handling, as well. Drivers have been a key factor in personal auto insurance rating. But going forward, once the driver doesn’t have a role, or the driver’s role is minimal, in getting around, the vehicle itself becomes more and more important in the underwriting and rating process. Then, in terms of claims handling, as we move toward automated vehicles having more of a role, it does lead to the question of whether the driver is at fault when there’s an accident or if the vehicle malfunctions. Where does liability come into play? Is it personal auto insurance, or does it become general liability insurance?”
Some manufacturers, including Volvo, will take ownership of liability for certain accidents involving their vehicles, she says. “But there will still be some questions about the driver and if the driver still has responsibility and was expected to take control,” Perfetto says. “There are still questions around what actions legislators will take in terms of the way drivers’ responsibility laws work today.”
There are some who wonder, too, if government agencies or legislation will deem autonomous vehicles so beneficial for society as a whole that they become more of a requirement than an option.
“Brokers, underwriters, agents and others need to think differently, even now, and need to be asking questions differently,” Perfetto says. “And some of that is related to ride sharing and what risks their drivers are assuming that weren’t there before. One of the things we see is that insureds may not always understand their policies, what their coverages are and what they’re covered for. With the whole sharing environment, people are putting themselves out there more, offering their cars for delivery services or taking on different roles. But they may not be thinking about the commercial risk they’re now undertaking. As a result, insurers need to be thinking about ways to improve their underwriting and renewal processes, their applications, to make sure they’re asking the right questions and appropriately evaluating the risk based on the driver, the vehicle and where that driver is operating.”
In this period of transition, Perfetto says, drivers also will have much to consider.
“Education is going to be key,” she says. “It’s not learning how to drive like you used to drive. It’s learning differently, or it’s not learning how to drive at all when you’re talking about the autonomous vehicles.”
In her own household, Perfetto has two vehicles, but only one has safety features such as a backup camera. “And when I drive the one without those features, I really have to pause and think what I’m doing.” Even with a backup camera, there’s still a need to look behind the vehicle. But it’s easy to depend on the technology—and take it for granted.
By 2018, backup cameras will be standard on all new cars sold in the U.S., thanks to a ruling from the National Highway Traffic Safety Administration, but many used cars will still be without.
“Having that mix, which a lot of households do, is already a learning curve,” she says. “People will need to learn how to drive and respond differently, given that mix of vehicles…. For a long time, the challenge is going to be this mixture of drivers, some having older vehicles, some with certain advancements in technology, and maybe some that are autonomous, with the driver still behind the wheel or, if there’s no wheel, still with the ability to take control.”
Forecasting the Future
Swiss Re’s Bongers, meanwhile, openly speaks to the impact driverless cars and advanced vehicle technology will have on insurance and other sectors. As automotive comprises roughly 42% of the total global primary property-casualty market—58% in emerging markets—it is an important field of study. The transition period to fully self-driving cars, Bongers says, will not be easy for insurers, and Swiss Re wants to be able to offer solutions to its clients.
The company has brought together a range of experts, including car manufacturers and specialists in technology, safety and legal issues, for automated vehicle conferences in Zurich and New York. Insurers are fascinated by the topic, he says, but there’s still much uncertainty about the coming impact.
His best forecast? First, he says, “We expect car ownership to change slowly over time. We estimate that it will take 10 years after the introduction of fully autonomous cars to reduce the amount of cars by approximately 10% to 20% in large cities in developed countries (estimated at 2035-2040). Studies show that 20 years after fully autonomous cars are introduced, the global fleet of cars could be reduced by 50%. We find this figure fairly optimistic. The change in ownership will be primarily visible in cities. From a risk perspective, we do not pay much attention to car ownership and car sharing because we believe … the amount of travelling will stay the same. For insurance companies, the amount of accidents is the key factor. And the amount of accidents does not depend on the amount of cars owned by people but on the amount of travel.”
Bongers does believe car sharing will gain popularity, though vehicle ownership will still be the preference for the next couple of decades. Cars will become extensions of our homes and offices as places where significant time is spent, with those inside watching movies, eating breakfast and checking emails. As for whether drivers will still need personal auto insurance, Bongers says that depends on local legislation. He expects some states and countries to proactively change third-party liability requirements, while others will wait to evaluate technologies before changing legislation.
“As long as there are moments that people have to drive the car themselves, there will be a need for personal auto insurance,” he says. While car manufacturers might assume third-party liability when the car is driving itself, he says, “even in this case, the owner of the vehicle may be the policyholder. The premiums are just paid by the manufacturer or fleet operator in the case of car sharing.”
Eventually, automated cars will cause fewer accidents. They’ll also help decrease congestion, especially as they’re able to re-route and/or drive closer together, Bongers says. And even Detroit can win in this new scenario, he projects, as long as car manufacturers begin shifting business models from focusing on new car sales to new forms of mobility.
All in all, it’s a lot to take in, and there’s too much to consider to let the unfolding technology drive its own future.
“When I first started looking at personal auto insurance, it wasn’t the most exciting line of business to work in,” Perfetto admits. “I don’t think I ever envisioned being in this position now, where there’s so much to think about and so many things to look at. It’s constantly changing, and there are new things we’re learning every day.”