On June 15, 2022, the National Highway Traffic Safety Administration (NHTSA) released its initial summary reporting on autonomous vehicle crashes collected through its Standing General Order. According to the NHTSA, in the last year, there were 130 reported crashes involving vehicles with levels 3-5 automated-driving systems. Although NHTSA’s recent summary does not specify what caused these accidents, we might anticipate two eventualities going forward: the number of reported crashes may increase as autonomous vehicle technology becomes more widely adopted, and the question of whether crashes were caused by the driving system or the driver will increasingly be a factor in determining liability. Against this backdrop, one might wonder how personal auto insurance will evolve to meet these changes.
By way of background, road vehicles come with varying degrees of automation technology. To differentiate between these varying degrees, the Society of Automotive Engineers defined different levels: “level zero” refers to fully manual vehicles, while “level one” refers to cars with basic driving assistance features (e.g., adaptive cruise control) and “level two” refers to cars with more advanced driving features (think Tesla Autopilot). Meanwhile, levels three and four refer to “conditional driving automation” and “high” automation, respectively. Level five, the highest defined level of autonomy, refers to “full” automation, where no driver is needed at all.
This article focuses on personal auto insurance for owners of level four autonomous vehicles, i.e., vehicles equipped with artificial intelligence systems capable of performing all driving tasks, but which also permit the driver to take full control if necessary or if desired.
The Road Traffic Act and the Compulsory Insurance Act, passed in Germany last year, provides something of a roadmap for what auto insurance may look like in the context of level four autonomous vehicles. Instead of focusing on drivers, the Act requires a “technical supervisor” who can deactivate the vehicle or take control of its driving maneuvers when necessary. Notably, the Autonomous Driving Act also requires the vehicle owner to maintain liability insurance for the technical supervisor. This seems to make sense – assuming an on-board technical supervisor is required to step in if something goes awry, then there should be insurance to cover any liability associated with the technical advisor’s failure to do so.
At this stage, it is largely unclear how technical supervisor insurance will be priced and how it would compare to premiums for traditional auto insurance. As consumers of car insurance are often aware, auto insurers rely on actual loss data in setting rates. Because highly autonomous vehicles are still in the testing phase, there likely will be a lack of loss data for auto insurers to reference, making it difficult to price the insurance. (As discussed in more detail below, it is at best uncertain whether auto manufacturers will freely share their testing data with auto insurers.) It is possible that technical supervisor insurance may demand higher premiums, the cost of which could be justified in part by additional coverages. For example, technical supervisor insurance could, hypothetically, offer coverage akin to travel insurance, covering the technical supervisor for accidents caused by his or her navigational decision-making, say, if the autonomous car gets caught in a blizzard during a mountain road trip. Such policies could also cover liability caused by the owner’s errors or omissions (e.g., if the car owner forgets to download a firmware update before operating). Conversely, however, it is possible that technical supervisor insurance could be less expensive than traditional auto insurance if it turns out there is less risk to insure.
Keep in mind that, in a level four autonomous vehicle, a technical supervisor who yearns for the feeling of being behind the wheel can still take full control of the car’s driving mechanics. Theoretically, when driving the car manually, the driver should be insured just as a traditional driver would be. Then, when the driver decides to take a break from the wheel and devote his or her attention elsewhere, that driver’s insurance should toggle to reflect differences in pricing, policy limits, additional coverages and/or other policy conditions that may apply.
Thus, a form of split usage-based insurance could provide the solution for personal auto insurance in the level four category. Indeed, companies like IMS are already partnering with carriers to enable usage-based, pay-per-mile, auto insurance. Split usage-based insurance simply takes this concept a step further to account for the split role between driver and technical supervisor.
This form of insurance could resemble frameworks that regulators across the country have implemented to set insurance coverage requirements for ride-hailing companies like Lyft. For example, the California Public Utilities Commission developed a framework divided into three “Periods” based on the driver’s status: Period 1 begins when a driver logs on to a ride-hailing application and becomes available to receive ride requests; Period 2 is the time between when driver accepts a ride request and passenger pickup; Period 3 is the time between passenger pickup and drop-off. The regulation requires the ride-hailing company to provide insurance at specified policy limits for each of the three Periods. In essence, personal automobile insurance for the driver/technical supervisor could function much the same way, with separate usage-based pricing, policy limits and other terms and conditions applicable based on whether the vehicle is being operated manually or autonomously at a given moment in time.
In order for split usage-based framework to function optimally, however, it would be critical for vehicle manufacturers to report to insurance carriers at least some limited vehicle status data (i.e., data that indicates whether, at a given moment in time, the vehicle was being operated autonomously or by the driver). But auto manufacturers may be reluctant to do so, because such data could be used against them in products liability lawsuits, either by an injured plaintiff or by the insurer in a subrogation proceeding. To address this, governments could standardize new requirements of the various stakeholders in a way that facilitates cooperation and trust, while being mindful of each one’s respective interests.
Eventually, the technology should reach a point where no onboard technical supervisor would be required. Perhaps this will be due to advances in artificial intelligence and computing power, the use of teleoperation, or both. Undoubtedly, much of the value to tomorrow’s consumers of highly autonomous vehicles will lie in the prospective ability to divert one’s attention away from the road (to catch up on work emails, read the morning news, play an online game) while zipping comfortably between A and B. Once the technology reaches that stage, there should be less third-party liability risk for the owner of the car, and perhaps then personal auto insurance will only be needed if and when a passenger occasionally decides to enable the manual functions of the vehicle. Until we reach that stage, a form of split-usage-based insurance seems like a viable solution for owners of level four autonomous vehicles.