This blog post examines the legal liability issues surrounding autonomous vehicles and the possibility of legal personality for artificial intelligence.
Have you ever imagined a car that drives itself and takes you to your destination while you sit still? Probably everyone has encountered driverless cars at least once in science fiction movies or novels. Now, this imagination is no longer a thing of the distant future. The day when we can actually see driverless cars on the road is fast approaching.
Research on driverless cars began decades ago. In the mid-1960s, an American company developed a car that could drive autonomously by detecting induced currents, and since then, numerous research institutes, universities, and private companies around the world have been competing to develop this technology. Even now, countless researchers are continuing their efforts to improve the performance of driverless cars and accelerate their commercialization.
One company stands out among this technological competition.
That company is Google, a leading US IT company. Google developed and unveiled a state-of-the-art driverless car based on its long-standing expertise in artificial intelligence, mapping, and sensor technology, instantly becoming an industry leader. Google’s announcement that the development of driverless cars was a realistic possibility made a huge impact around the world. However, despite high expectations, autonomous vehicles face many challenges in the real world.
In particular, between 2023 and 2024, there have been a series of reports of accidents involving self-driving cars in the US, both large and small, sparking social controversy. Waymo, Google’s subsidiary in charge of developing self-driving cars, has also experienced minor collisions during test drives in some cities, reigniting concerns about the judgment and safety of these vehicles.
In particular, some accidents were attributed to errors in the AI judgment of the vehicles themselves, rather than simple external factors, sparking debate over the legal responsibility of autonomous vehicles. This is seen as an example of how, despite the high level of autonomy achieved by autonomous driving technology, it still has limitations that differ from human intuitive judgment.
What attracted the most attention was the question of who was responsible for the accident. When a car is driven by artificial intelligence rather than a human driver, who should be held legally responsible in the event of an accident? A fundamental question was raised as to whether the artificial intelligence computer that drives an unmanned car can be considered a “driver,” that is, a person.
If so, can autonomous vehicles be recognized as “human beings” under the current legal system in Korea? The term “human” here refers not simply to a living being that performs physiological activities, but to a legal entity that is subject to legal rights and obligations. Korea’s civil law defines legal entities through the concept of “natural persons,” and only when a natural person has a certain level of decision-making and behavioral capacity can they be held legally responsible and obligated.
Decision-making ability refers to the ability to recognize and judge situations, while capacity to act refers to the ability to independently perform valid legal acts based on such judgments. Therefore, whether or not the AI computer of an unmanned vehicle has such decision-making ability and capacity to act is the key criterion for determining whether it should be granted legal personality.
In order to determine these abilities, it is necessary to examine what judgments and actions unmanned vehicles can make on their own in actual driving situations. The key is whether they are capable of setting their own destinations, selecting the optimal route, and actively responding to various variables that may arise during driving. It is also necessary to confirm that they can recognize traffic signals and road signs, adjust their distance from other vehicles and pedestrians, and respond safely to unexpected situations.
The way in which driverless cars select routes to their destinations is similar to the navigation systems we commonly use. They determine their current location using a satellite navigation system and then use a search algorithm to set the shortest route. However, while general navigation systems merely assist the driver, driverless cars perform the driving itself, so they require much higher accuracy and reliability.
To this end, correction technology has been applied to driverless cars. Compared to existing satellite navigation systems, which have an average error range of about 10 meters, the latest technology has succeeded in reducing the error to within 0.1 meters. This enables driverless cars to determine their location more accurately and maintain a stable route.
However, no matter how advanced the technology is, there are still fundamental limitations to driverless cars. This is because they can only move within the map information stored in their memory. In other words, unpaved roads that have not been entered in advance and roads that have been temporarily changed due to construction are excluded from route search, and since these changes are not recognized in real time, there is a possibility of errors occurring while driving.
In addition, driverless cars are equipped with various sensors to recognize the external environment in real time while driving. Among these, the most essential device is the LiDAR sensor installed on the roof of the car. This sensor rotates 360 degrees and emits lasers, detecting signals reflected from surrounding objects and recognizing hundreds of objects simultaneously. In addition, camera sensors and radar sensors are also installed to accurately detect traffic lights, road signs, and the speed and distance of surrounding vehicles. Thanks to
these technologies, today’s driverless cars can go beyond simple object recognition to distinguish the number of people standing at a crosswalk and even recognize the gestures of people waving from car windows. The various information collected by the sensors is processed in real time and converted into a three-dimensional map, which the vehicle uses to determine whether to slow down, turn, or stop.
The decision-making ability of driverless cars is based on experience data obtained from actual road driving. This is why driverless cars become “smarter” the more they drive. The more data they learn, the better they become at handling various situations, and the more sophisticated their algorithms become.
However, this learning ability, which is also a strength of driverless cars, reveals clear limitations. Driverless cars can only respond properly to situations they have experienced. In other words, they are unlikely to make appropriate decisions in situations that are not stored in their memory or that have never occurred before. For example, in unpredictable situations such as animals crossing the road, sudden heavy rain or snow, or icy roads, the judgment of autonomous vehicles will inevitably reach its limits.
As such, autonomous vehicles clearly have the ability to set their own routes, recognize driving conditions, and make decisions, unlike conventional vehicles. This can be seen as evidence that they have a certain degree of decision-making and action capabilities.
However, they still lack the ability to adapt to environments that are not stored in their memory, and there is still a long way to go before they can achieve complete autonomy. For these reasons, it can be said that there are currently many shortcomings in granting legal personality, or legal responsibility equivalent to that of humans, to driverless cars. Nevertheless, technology is advancing rapidly, and it is possible that the limitations that are currently apparent will gradually be overcome.
If, in the future, driverless cars reach a level where they can recognize situations on their own and make human-like judgments even in exceptional circumstances, then the issue of granting legal status to artificial intelligence can be seriously discussed. We look forward to the day when driverless cars will be recognized as “human beings” in the legal sense.
