Expert Risk Articles

Autonomous Driving: The Road Ahead

With fewer accidents and more complex claims, driverless cars will require a very different approach to insurance.

An Uber robot car takes you to the train station and when the train arrives an autonomous bus or a driverless taxi is waiting to transport you to your final destination. All organized from a smart phone, all automated and seamlessly coordinated.

As Jacob Fuest, Head of Automotive Innovation Center, Allianz Global Automotive tells Global Risk Dialogue, this is the future of transport. It will be many years before such a scenario becomes reality, but a transformation is underway, he believes.

Three clear trends are emerging, says Fuest – that of connected cars, increased automation and shared mobility. New cars will increasingly feature connected and driver assistance technology, while a host of technology companies and car manufacturers are now developing and testing autonomous vehicles for private and fleet use.

Ethical and liability debate

However, a smooth ride for driverless cars is not guaranteed. According to Carsten Krieglstein, Head of Liability at AGCS Central & Eastern Europe, the introduction of driverless cars will require changes to existing laws around liability, as well as an ethical debate.

Under the relevant national laws, which are shaped by the 1968 Vienna Convention on Road Traffic, the driver of the vehicle is responsible for damage and personal injury to third parties in an accident. Although the Vienna Convention was amended in March to allow driver assistance technologies, the driver must remain in control and is responsible for any damage.


As autonomous vehicles hit the streets, risks will shift from human-errors to product-errors. [Photo: Fotalia]

“An important question for the future will be whether a car is driven, not by a person but by a machine,” says Krieglstein. “For fully automated driving, the legal environment would need to allow a vehicle to be liable for an accident. This is a
major obstacle that needs to be overcome,” he explains.

A potentially more taxing challenge for driverless cars will be how machines perform when faced with ethical decisions. For example, if a child runs out into the road, a vehicle may have to choose between hitting the pedestrian or swerving at the risk of hitting a cyclist.

“Technology is not the problem. But we do need to recognize that difficult ethical decisions will need to be made by machines,” says Krieglstein.

Advanced driver assistance systems

True driverless cars may be some way off, but advanced driver assistance systems (ADAS), which range from automatic parking to collision avoidance and
autonomous cruise control systems, are already being incorporated into many brands of cars.

These automated technologies will pose their own risks, such as an increased risk of cyber-attack or systems outage. The technology is also still very much in a test phase.

At present, such technology is best suited to driving on highways, and has yet to be proven in more challenging driving conditions, such as city centers, explains Krieglstein. So drivers will need to remain in control of their vehicle for many years to come, he predicts.

And when true driverless cars do take to the roads in significant numbers, they will have to co-exist with non-automated traffic, including cyclists, motorbikes, as well as other, and older, cars, he adds.

Human versus auto

The case for greater automation is compelling – cars should be more efficient and more environmentally-friendly, and most significantly, they should be safer. It is estimated that human error contributes to more than 90% of road accidents, so the expectation is that accident rates would fall significantly with driverless cars.

“Systems make fewer mistakes than humans and we already have witnessed a decline in claims frequency for vehicles using advanced driver assisted systems,” says Fuest.

However, while autonomous technology will help reduce accident rates, accidents will continue to happen. For example, animals or pedestrians will still unexpectedly step out into the road, while, of course, technology is not infallible.


Extreme weather conditions, like snow or rain, could interfere with sensors while driving conditions could push technology beyond its limits. And there will be a question over the ability of technology to perform reliably and effectively over the life span of the vehicle. In addition, humans are behind the technology, software and algorithms, so there will still be opportunities for mistakes.

There have been fatal accidents involving Tesla cars that allegedly involve the vehicles’ advanced cruise control systems. For example, the preliminary report into a 2016 Tesla crash in Florida by the National Highway Traffic Safety Administration (NHTSA) noted that Tesla’s “autopilot” had failed to break for a white-colored truck against a brightly lit sky.

Accidents of the future

Causes of accidents in the future are likely to look very different and liability will be more complex, according to Fuest and Krieglstein. While today, liability lies mostly with the driver, in the future manufacturers, software developers and fleet operators will assume much more liability, they explain.

“With fewer accidents, motor insurance will not have the same impact in the future but there will be an increasing shift to product liability for motor manufacturers and product recall. Cars are becoming more complex and any technical fault could result in a major recall,” says Krieglstein.

A recent study by insurance broker Aon predicts that autonomous vehicles could reduce US motor insurance premiums by more than 40% by 2050[i]. Its estimate is based on an 81% reduction in claims frequency, and an increase in claims severity due to sensor costs and an increased cost of handling product liability claims.

However, as long as there are serious accidents, there will continue to be a need for some form of compulsory third party liability insurance. And for accidents involving autonomous vehicles, this could see the insurer compensate third parties and then pursue the manufacturer if they are at fault.

“In the future we will see changes in how motor insurers assess their risks. However, the risk for the insurer won’t disappear completely, instead shifting from the ‘human error’ to the ‘product error’ category,” says Krieglstein.

Strict liability combined with insurance will remain the “optimal solution” for resolving third party liability, believes Fuest.

“In many ways, insurance will outwardly continue to appear as it does today and it is likely to remain mandatory for vehicles on the road for the foreseeable future. However, there will be different processes going on in the background and it will be delivered in a wider variety of ways,” he says.

Implications for insurance

Automation will have further implications for insurance, according to Fuest and Krieglstein. For example, technology is likely to contribute towards a decline in car ownership, in favor of motor fleets, car-sharing and driverless taxis. This could see insurers move away from providing millions of single annual motor insurance policies to drivers, instead providing large policies purchased by manufacturers and fleet owners and operators.

Car manufacturers are already looking to offer broader “mobility services” that bundle running costs and insurance into a single fee. The shift to product liability will require insurers to develop technical expertise and not rely on historic data and driver profiling for pricing. Allianz has already started building teams of engineers with experience in automotive and driverless technology.

“Our automotive clients are working on developing driverless and driver assistance technology, and are looking for the support of insurers. Allianz is working with vehicle manufacturers and suppliers, discussing the potential risks and developing future coverages,” says Krieglstein.


Allianz has been developing insurance products in a number of areas. For example, it offers discounted motor insurance for drivers with cars that already include driver assistance technology (see section in right-hand column). Allianz is also insuring driverless technology during testing as well as autonomous vehicles – in the US and the Netherlands Allianz insures EasyMile driverless buses. Meanwhile, together with auto giants BMW and Toyota, it has also recently joined forces with autonomous vehicle technology company Nauto. Under this agreement it will license data and technologies in order to better understand how drivers and vehicles behave and perform.



[i] Global Insurance Market Opportunities (GIMO) report, 2016, Aon
[ii] Automotive Manager 2015, Oliver Wyman