ÈȵãºÚÁÏ

While the media, ethicists and perhaps the public like to ponder whether a self-driving car in a crisis should be programmed to hit a group of nuns crossing the street or hit a bus full of schoolkids, the drive toward autonomous driving is unlikely to be slowed by such moral dilemmas, insurance executives were told recently.

That’s because while scenarios involving split second decisions and catastrophic consequences are worth considering, the odds of some of them happening are very low whereas the odds of automated driving saving many lives are very high, Joe Schneider, managing director, KPMG Corporate Finance, suggested before attendees at the Super Regional P/C Insurer Conference in Lake Geneva, Wis.

“My personal viewpoint is that in many ways it’s kind of a thing that the media latches onto because it’s a very interesting hypothetical question,” said Schneider, when asked about programming ethical choices. “I don’t know if in reality, there’s going to be many nuns crossing the street and school bus situations. With that said, it is something to certainly consider.”

(As Schneider pointed out, MIT is one of the places where academics are studying moral choices. MIT scientists have developed a platform called , which allows users to judge and decide what they would do when faced with driverless car moral dilemmas such as killing two passengers or five pedestrians.)

The point stressed by Schneider and other panelists at the Super Regional conference is that automated driving will be fueled more by safety than by anything else. Many people will appreciate that driverless cars free-up time, some firms will benefit economically by cutting the costs of employing drivers, and others who have not been able to drive will welcome their new freedom to travel. But the paramount promise of driverless cars is that they will save lives. They will do so by eliminating human error, which is responsible for most traffic fatalities in the U.S. every year.

“The important point with the almost 40,000 fatalities is that over 90 percent of those are attributable in some way to human or driver error, whether it’s distracted driving or whatever,” said panelist Peter Rafferty, who is manager of the Transportation Systems Management and Operations (TSM&O) unit at the University of Wisconsin and involved in the university’s U.S. government-approved testing site for driverless vehicles.

“The automated vehicles… they don’t fall asleep, they don’t get drunk, they see all around them. They react much faster,” said Rafferty. “So that’s where big safety gains are going to be happening.”

KPMG’s Schneider brought a copy of his firm’s most recent research into what all this means for the auto insurance industry. In a 2015 report, KPMG said the market would shrink by as much as 60 percent by 2040. However, according to KPMG’s updated study, the auto insurance sector will dwindle by more than 70 percent or $137 billion by 2050 due not only to autonomous vehicle technology but also a rise in on-demand transportation and a shifting of liability to manufacturers.

According to many researchers, the transition from mostly human-controlled to human-free motoring will happen over the next 25 years. The road will likely be marked by regulatory bumps and a “chaotic middle” as KPMG calls it, or a “messy middle” as Rafferty characterizes it, where vehicles with varying levels of autonomy and human involvement are on the roads.

Automation Levels

The Society of Automotive Engineers (SAE) has developed widely-accepted for the various levels of automation in driving, which Rafferty highlighted. There are six levels starting with zero. In levels 0 to 2, the human driver monitors the driving environment. In levels 3 to 5, the assisted driving system monitors the driving environment:

• 0. No Automation— human drivers perform all aspects of driving even when there are warning or intervention systems.
• 1. Driver Assistance — driving assistance systems of either steering or acceleration/deceleration. Example: automatic braking systems.
• 2. Partial Automation— one or more driving assistance systems of both steering and acceleration/deceleration. Example: Tesla Autopilot
• 3. Conditional Automation —automated driving system performs all aspects of driving with expectation that human will respond appropriately if requested to intervene.
• 4. High Automation— automated driving system performs all aspects of driving even if human does not respond appropriately to request to intervene requested to intervene.
• 5. Full Automation— automated driving system performs all aspects of driving under all roadway and environmental conditions.

“At level zero, one, and two, the driver is supposed to always be paying attention and ready to take over control of the vehicle if something goes wrong, like the Tesla auto-pilot. Levels three, four, and five that’s not the case. If something goes wrong, the vehicle is able to go into some sort of safe condition. That’s the important distinction here,” Rafferty said.

The “messy middle” will occur between the level two and three automation period of partially automated vehicles, where the driver should be paying attention but may be distracted. In later levels, eventually the driver will not be required to pay attention and the cars will shift into a safe mode of operation if a dangerous situation arises.

Timeline

Experts have offered different adoption timelines; Rafferty’s calls for the transition being largely completed by 2040.

“It’s gonna’ be years or even decades yet before we really get the full automation that can operate everywhere. The vehicle can get out of your garage, go through a neighborhood across town, pick up a kid, what not. That’s still a long way away. Automated vehicles are here, yes. But we’re just at the beginning of a rather long transition period in my opinion,” he said. “This is just my impression of things.”

Rafferty expects that by 2020, there will be plenty of level three cars on the roads. By 2025, “shared mobility” fleets using firms like Uber and Lyft will have caught on. By 2030, highly-automated level four vehicles will start appearing and by 2030, human drivers will be banned on certain roads and in urban areas. By 2040, the transition is going to be largely complete, Rafferty believes.

“From there to 2100 it’s going to be like the shift that we just had 100 years ago from the horse to the gas-powered automobile. It’ll be that dramatic. This is really going to be a change. Not that these things will be outlawed. It’s just that, like today, you can’t ride your horse around in a lot of places. You probably won’t be able to drive a vehicle around that can kill people.”

In outlining his timeline to 2040, Rafferty offered a bit of career advice that elicited some nervous laughter. “What you can do if you’re in the auto insurance business, you can just kind of find where your retirement is on this,” he half-jokingly told the audience of P/C insurance executives.”

KPMG has offered its own timeline. Over the next two or three years, consumers will begin experiencing the safety advantages of new technologies and attitudes will shift towards acceptance of autonomous driving. Also the first autonomous cars will be on the roads. On-demand transport and car-sharing will continue to expand. By 2024, the majority of travel within cities and surrounding suburbs will be on-demand rather than with a personal vehicle, and by 2035 on-demand will be the norm in transportation, according to KPMG’s projections.

ÈȵãºÚÁÏ Pros

Today’s insurance executives should be educating themselves on the technology and possible timelines because the transition will be happening on their watch, according to panelists.

“This is not George Jetson stuff,” KPMG’s Schneider said. “I’ve been to more car shows than I care to remember in the last two years. This stuff is real. It’s moving more quickly.”

He urged the Super Regional executives to prepare.

“I think the first step for preparing an organization is just for people to think that this isn’t something that will come and will have impact in 10 or 12 years when someone’s in retirement,” Schneider said. “It’s more a situation of, ‘Hey let’s understand the changing landscape of auto, as well as then what are the fundamental impacts on our organization and go from there.”

One unknown is what regulators will do to influence the timeline. Rafferty noted that various states have rules on the books or in progress regarding testing of autonomous vehicles and Congress is looking at ways to allow manufacturers to bypass states to speed up development.

The U.S. Department of Transportation last year invited researchers to be part of a virtual network of automated vehicle proving grounds. Of the 10 sites selected, Rafferty’s University of Wisconsin was one.

His own state also has its own regulatory group—cleverly titled a “steering committee,” he noted —looking at how to proceed on public roads.

Although autonomous vehicles are expected to become very popular, it’s also expected that some people will still drive.

Schneider recalled a recent global corporate finance meeting in New York where he gave a presentation on the topic to insurance professionals. There was a gentleman in the back of the room from Milan who begged to differ that driving will become extinct. In his Italian accent, he insisted, “People will always be driving Ferraris. You’re not gonna’ make that car autonomous because people are always driving Ferraris.”

The Super Regional P/C ÈȵãºÚÁÏ Conference was sponsored by actuarial consulting firm Demotech Inc. and Wells Media’s ÈȵãºÚÁÏ Journal and Carrier Management.

Topics USA Auto InsurTech Property Casualty