This blog post takes a balanced look at the social benefits of driverless car technology and the ethical and technical issues behind it.
Recently, Google and other companies have been actively researching driverless car systems. Driverless cars are cars that drive themselves without a human driver. If this technology is realized, computers will replace drivers, and all city roads will be automated like factories, bringing revolutionary changes to society. Google has already successfully completed tests of driverless cars on highways and in city centers, and as of 2023, Waymo autonomous vehicles have begun paid services for the general public in Phoenix and San Francisco. Waymo CEO Tekedra Mawakana said, “Our goal is to build a safer and more efficient transportation system.”
In response, the RAND Institute, a non-profit research organization that serves as a defense and administrative advisory body to the US government, also published research findings supporting this view, stating that the benefits of driverless car systems to society will outweigh the side effects. According to an official article by the RAND Institute, James Anderson, a senior researcher at the institute, found that driverless cars have significant social benefits, such as eliminating car accidents through collision control, improving mobility, reducing car ownership, allowing cars to be called anytime and anywhere, and increasing fuel efficiency and reducing fuel consumption.
He gave a positive assessment of these research results, and RAND recommended that policymakers actively prepare for the introduction of driverless car systems by readjusting relevant policies, including traffic regulations. As a result of these positive reviews and lobbying by companies, several states have already passed laws on driverless cars that are now in effect.
In June 2011, the state of Nevada became the first to pass a law allowing driverless cars, and in May of the following year, Google’s driverless Toyota Prius was issued a license. Subsequently, in April 2012, Florida allowed driverless cars to be tested on public roads, and in October of the same year, California and Michigan passed laws allowing the limited operation of driverless cars. In the 2020s, Arizona and Texas passed laws allowing the commercial operation of driverless cars, and autonomous vehicle companies such as Waymo are now actively operating in these areas.
However, at this point in time, when the introduction of driverless car systems is spreading, it is necessary to reexamine whether the benefits of this technology outweigh its side effects. If the currently incomplete technology for driverless cars is sufficiently developed and reaches a level ideal for actual implementation, many benefits can be gained, such as the elimination of vehicle collisions and improved urban space utilization, as shown in the RAND study. However, even if technological development proceeds ideally, there are still issues that cannot be resolved.
First, who should manage citizens’ mobility information and how? When driverless cars are in operation, their locations will be collected and managed in real time, as is currently the case with public transportation vehicles such as subways, trains, and buses. Current public transportation operates according to a fixed schedule, but driverless car systems assign appropriate vehicles to requests from users and drive them along the requested routes, so it is necessary to know the exact locations of the vehicles in order to assign the appropriate vehicle based on the location of the user who made the request. Someone has to make decisions based on this location information, so there must be a central control system that can refer to all location information. In this case, the control system administrator will have access to this information. In other words, all movement information of citizens who use driverless cars will be disclosed to a small number of responsible persons and stored and managed in the system. This in itself is a serious invasion of privacy and carries the inevitable risk of hacking. Hacking and security are relative issues, depending on which side discovers the loopholes first, so as long as information is collected centrally, the risk of hacking will inevitably exist and cannot be fundamentally resolved.
Second, there is the problem that all connected vehicles may be damaged due to a malfunction of the central system. This problem also arises from the nature of the structure, in which terminal devices are controlled centrally. Unmanned vehicles decide what movements to make based on the surrounding conditions recognized by their sensors, but they do not always make decisions on their own; sometimes they must move according to commands from a central control system. For example, vehicles must be able to be controlled by decisions made centrally in order to move to areas where there are not enough vehicles, undergo maintenance, or respond to emergencies such as natural disasters. The ability to control each device from a central location means that even a small error in the central control system can cause all devices receiving commands from the center to be adjusted incorrectly, potentially causing numerous vehicles, or in the worst case, all connected driverless cars, to malfunction. Even with advances in technology, it is not easy to guarantee the integrity of a central control system because it is software created by humans. There is also the risk of hacking. If a hacker infiltrates the system with malicious intent and causes the central system to malfunction, the damage will spread to all connected devices, causing device damage, inconvenience due to traffic paralysis, and even loss of life due to malfunction. Of course, even devices such as subways and airplanes can malfunction, but the difference is that subways and airplanes are operated at a low level of automation and are always supervised by human managers who make higher-level decisions. In contrast, unmanned vehicles have no human involvement in the control of the device, so there is a risk that a system malfunction could directly lead to a device malfunction.
Finally, there is the issue that decisions made in unexpected situations are made entirely by the system designer. The elimination of collisions in driverless car systems means that collisions between driverless vehicles controlled by a central authority will disappear, but it does not mean that all potential accidents can be prevented completely. In real-life situations, unexpected situations that cannot be predicted by the system can occur at any time. For example, suppose a pedestrian suddenly crosses the road against the traffic signal. In this case, the driver will try to avoid colliding with the pedestrian, but if the driver swerves, the vehicle may fall off a cliff or collide with an oncoming vehicle, and damage will inevitably occur. In extreme cases, the driver may be faced with a situation where they must weigh the lives of the pedestrian and other drivers against their own. It is difficult to choose either option, but the important thing is that the parties involved in the accident make their own decisions on how to respond. However, this is not the case with driverless cars. All of the vehicle’s responses to situations are based on rules set in advance by the system designer, so decisions in unexpected situations are also made by the system designer. For example, if a person jumps out onto a road that is blocked by a wall on one side and a cliff on the other, and the designer has set a rule that the vehicle should brake but not swerve, the driver will have no choice but to hit the person, regardless of whether it is a child or an elderly person. In such a situation, only the driver can decide whether to swerve, and no one else has the right to apply a uniform rule to that decision. If designers create rules that always protect pedestrians, who would willingly ride in a vehicle that sacrifices itself in every unexpected situation? The opposite is also true. Creating such rules is tantamount to prioritizing potential loss of life, which must never happen.
The social benefits announced in the RAND study, such as fuel savings, reduced traffic congestion, and more efficient use of space, are on a different level from issues related to human casualties, so it is premature to conclude that the benefits far outweigh the side effects based on these social benefits. The introduction of driverless cars also has the advantage of reducing human casualties from traffic accidents, so the benefits of introducing driverless cars cannot be converted solely into costs and convenience. However, even if human casualties are reduced, if the direction of human casualties is determined by rules set by humans, such a change cannot be considered desirable. Although accidents are more likely to occur when humans are driving, at least drivers decide their own movements, and the extent of damage caused by accidents is determined solely by the actions of the parties involved, not by others. Of course, the choices made by the parties involved affect the lives of other parties involved, so they cannot take full responsibility for their choices, but this is completely different from having everything determined by a third party who is completely unrelated to the situation.
However, preventing the introduction of the system itself is not a wise choice, as it means throwing away the innovative benefits that new technology will bring. Furthermore, historically, regardless of how much conflict there has been in the process, modern humanity has always chosen to advance technology, and machines have replaced humans as a result. Therefore, simply preventing the introduction of the system would be nothing more than a temporary measure. Instead, we should strive to design alternative systems that can partially utilize the benefits of this technology and move toward technological innovation without side effects. For example, if limited autonomous driving functions are provided in situations where the entry of objects other than vehicles can be well controlled, such as on expressways, it will be possible to provide drivers with rest time and prevent traffic congestion caused by accidents while preventing side effects caused by central control or the entry of external factors. If unmanned vehicle mode is activated in parking facilities, drivers will not have to worry about finding a parking space or parking their cars, and systematic management will greatly improve space utilization efficiency, avoiding side effects and maximizing benefits.
