인공지능 로봇에 관한 형사책임 - 인공지능형법에 의한 형사처벌가능성을 중심으로 -

According to a report by the U.S. Road Traffic Safety Administration, when looking at the development stage of self-driving vehicles, Level 1 means cruise function and Level 2 means integrated automation vehicles. Level 3 vehicle is a partially autonomous vehicle and can be manually switched if necessary, and level 4 means a completely unmanned autonomous vehicle. This paper mainly deals with level 3, which refers to a vehicle designed to allow the driver to fully leave the driving to the autonomous driving technology of the car in autonomous driving mode and allow the driver to drive the car himself when switching to general driving mode. Self-driving cars can be seen as different in the strict sense from unmanned cars that move without people on board, but in reality, they are used interchangeably in terms of terms. In the case of artificial intelligence self-driving vehicles, commercialization is expected soon, and in this case, it is skeptical that it is meaningful to impose penalties on self-driving vehicles in the event of an accident. It is unlikely that it will be meaningful to impose capital punishment and prison sentences equivalent to natural persons on self-driving vehicles. This is because disassembling self-driving vehicles or storing self-driving vehicles in a certain place seems to have no criminal legal meaning at the moment. It is also questionable whether resetting the program of artificial intelligence has a punitive meaning close to the death penalty for humans. Artificial intelligence will not be able to be fined. This is because artificial intelligence robots cannot accumulate assets, and artificial intelligence robots will not open deposit accounts and make financial transactions like humans. Therefore, it is not easy to imagine at the current stage of holding human-equivalent criminal liability. Because it learns and develops on its own, the progress of artificial intelligence robots may actually be made faster than humans imagine. As mentioned for a while in the beginning, it should also be kept in mind that the timing of artificial intelligence singularities may come. However, so far, taking criminal responsibility for artificial intelligence is unlikely to be possible by the current criminal concept until it reaches the level of an artificial intelligence robot that can make moral judgments. Artificial intelligence robots will not be able to put artificial intelligence robots in targets where punishment norms are applied to categories different from human categories. This is because it will not be easy to recognize the behavioral concept of artificial intelligence robots with criminal legal meaning.