NorthAll legal barriers will need to be overcome to make autonomous vehicles the “next big deal” in the automotive sector, he says. Hazep Boryresearch director of the GlobalData thematic department.
Fully autonomous vehicles (AV), also known as driverless vehicles, have become a major aspiration in recent years as Tesla’s popularity with its autopilot capabilities and significant advances in artificial intelligence (AI) technology in various fields – as well as early overly promising various vendors , including Tesla itself, Waymo / Google and Uber (now Aurora).
Indeed, GlobalData has identified autonomous vehicles as one of four devastating mega-trends that are currently affecting the automotive industry.
However, we are still far from what the average person would understand as a driverless car, which the Society of Automotive Engineers (SAE) calls “level 5”, or full automation of control in any environment without passenger involvement.
In fact, there are no Level 4 or Level 5 cars on the market, and even Tesla cars with autopilot function are currently classified as Level 2 only because the passenger must remain attentive and take control if necessary.
This contrasts with statements by Tesla CEO Ilona Maska in mid-2020 that the company is “very close” to achieving Level 5 autonomous driving technology, as well as with forecasts from companies including General Motors, Google Waymo, Toyota and Honda that they will work on their own. driving on the roads until 2020.
However, apart from the problems with the driverless technology itself, those related to regulation (and especially insurance) are no less scary.
It can be argued that the lack of progress in insurance for self-driving remains a major stumbling block for its endorsement in the industry, and may ultimately slow adoption when technology finally meets its expectations.
The latest GlobalData thematic report on autonomous vehicles (AV) emphasizes that there are important and complex regulatory, legal and insurance issues that need to be addressed as vehicle autonomy evolves, especially when partially autonomous vehicles require occasional oversight. man.
Interestingly, while regulators are gradually working to develop a structure that will allow AVs to operate in increasingly demanding scenarios, insurance and reinsurance companies are relatively calm.
To understand why it is important to consider the basics of the insurance industry. The insurance contract or policy stipulates that the policyholder assumes a guaranteed, known and relatively small loss in the form of a premium payment to the insurer in exchange for the insurer’s promise to compensate the policyholder in the event of covered loss.
The insurer makes a profit in this business in the long run as long as the total amount of premiums is higher than the compensation payments due to unprofitable events that occur with some insurers. Therefore, insurers need a large amount of historical data from which they can calculate expected accidents and losses based on historical averages and probability distributions. In addition, losses must be rare and accidental, so diversification of risks plays in favor of the insurer.
Given this, there are two significant problems in autonomous vehicle insurance. First, there is very limited historical data on accidents involving cars of this type, so historical averages and probability distributions cannot be relied upon to predict future accidents, and there is no great depth in data to analyze age, gender, education, or geography that could afford the proper price of risk.
Second, and this is very important, accidents involving autonomous vehicles are likely to have a high correlation. While they may be due to ambiguous satellite coverage, firewall failures, or corrupted software downloads – all quite random events – there is a risk that they may be due to faulty software vendors or cyberattacks.
Reality bites: Autonomous car
There were many reports of accidents involving Tesla cars when autopilot was activated, and Waymo also had several accidents during the tests.
As an example, in early 2022, Tesla had to recall 53,822 vehicles from the United States with Full Self-Driving (Beta) software, which allowed some models to make “rolling stops” and not stop at some intersections, which posed a safety risk. . How many accidents occurred due to this malfunction before the recall is unknown, but that number was to occur by decision of the U.S. National Highway Traffic Safety Administration (NHTSA).
In the event of faulty software or cyber attacks, the damage associated with the crash will be strongly correlated as they will occur on all vehicles of this manufacturer, especially if they seek to support all customers on the latest software. Algorithms do not make random mistakes like human drivers that are emotional and affect personal circumstances, but rather determined.
From an insurance company’s perspective, this is a big problem because correlation risk leads to greater variance in expected losses, which will require higher insurance premiums. In addition, with very limited historical data on autonomous driving, there are no proven and reliable actuarial tables to compensate for injuries and deaths in driverless accidents. Thus, it is very difficult to “assess risk”, which explains why insurers and reinsurers deliberately take a very cautious approach.
For example, in the UK, the Association of British Insurers (ABI) is currently conducting various pilot programs with insurers such as Direct Line Group (DLG), AXA, XL Catlin and RSA, but on its homepage it clearly shows: “However, the insurance industry has realized that drivers cannot be given unrealistic expectations – in the foreseeable future we do not expect these cars to have sufficient backup features to allow drivers to be completely off the road.”
The risk of algorithm failure is quite unique. Although at first this may seem comparable to natural disasters such as earthquakes and floods, which also affect many policyholders at once, there are important differences. In the event of a natural disaster, insurance companies diversify by spreading their insurers geographically, so when an earthquake, say, in Japan, they do have many insurers in other regions whose premiums help cover huge compensations belonging to the Japanese. However, how can an insurance company diversify the risk of algorithm failure if all users around the world run the same software?
Despite the obstacles, there is progress
Last year, global reinsurance giant Swiss Re and Chinese technology giant Baidu announced a partnership to promote insurance and risk management for autonomous driving and autonomous vehicles. However, the first product they had in mind was not too ambitious stand-alone parking insurance, and other than that, they committed themselves only to focusing on research into risk management and insurance innovation for automated car management products.
An alternative could be the driverless car manufacturers themselves, which insure their products, as Rivian does, which may provide a good match between insurance risk and the cost of software quality assurance. However, good risk management practices will require them to reinsure part of that risk, which will again require the involvement of large reinsurance providers.
However, problems with autonomous driving insurance should not be seen as an indicative stop, but as an argument to restrain expectations. There is still a long way to go until 5-level cars do appear on the roads, and secondly, there is an argument that there will generally be far fewer accidents if most vehicles are autonomous. And, not surprisingly, several turkey startups are addressing this complex issue, such as Koop Technologies, Avinew and Trov, and these are just a few.
The remarketing sector is geared towards the use of connected vehicle technology: epyx