자율주행차의 발전단계로 본 운전자와 인공지능의 주의의무의 변화와 규범적 판단능력의 사전프로그래밍 필요성

As technologies regarding autonomous driving advance from Level 2 to Level 3 automation, the role of a human in operating a car is diminishing and now expected to be replaced by an artificial intelligence(“AI”) embedded in an autonomous vehicle(“AV”). A question may arise as to the level of the duty of care that the replacing AI has to pay. This article argues that to the extent an AV takes over the role of a human driver in car operation the AI should play the same role as a human driver, and thus the level of the duty of care that an AI has to pay should be as high as the level that is reasonably expected of a human driver who controls his driving condition. This level of duty of care reasonably expected of an AI should also be reflected as a legal standard in the Automobile Safety Standard so that auto makers will be required to comply with in self-certifying their automobiles production. A more complex question may arise in Level 4 AVs, in particular, in respect of ethical issues, Because a steering wheel and a brake pedal disappear in Level 4 AVs, there is no room for a human driver to interfere in driving such a Level 4 AV. That means where the AV faces a conflicting duties whether legal or ethical duties, the AI in the AVs rather than a human have to decide whate to do by itself. Consequently, automakers should ex ante consider certain situations where its duties may conflict, and determine and programme what to do in advance. This requirement should also be implemented in the Automobile Safety Standard.