인공지능과 도덕적 기계 － 칸트적 모델과 흄적 모델

초록

Artificial intelligence is a system of information and rules, and a formal system of coding, programming, and deep learning algorithms. Deep Learning 2.0 version is an algorithmic operating system based on artificial neural network structure, and it learns big data on its own and handles things without being blocked by precise and accurate calculations and quick judgments. It seems like an ideal being that combines theoretical and practicality into one. In the case of humans, there are capable and bad people, but artificial intelligence, which chooses only morally right things, can be seen as artificial moral agents or moral machines. However, we generally apply the term 'moral' to those who are free to choose to think and judge for themselves. Human beings are considered to have this sense of moral autonomy. According to this, moral machines are impossible in principle. Thus, artificial moral agents should be viewed as beings with only engineering autonomy, which is nothing more than the automation of automatic machines. However, there is also a view that this is only reasonable if a particular moral view of moral autonomy is adopted. This article attempts to design the possible models of the moral machine in the position that there may be a variety of perspectives understanding ‘moral’.

키워드