인간 편향성과 인공지능의 교차

It is argued that the decision-making of artificial intelligence (AI) is problematic because it reflects human bias like a mirror, and that there is a need for regulation. However, even if we accept the premise of this argument, it seems that the conclusion cannot be logically drawn. This is because there is no reason to create a new regulation system if there is no change in the total amount of harm. And since the root of all problems is human beings, it is likely that we should pay attention to the existing regulatory system focusing on human bias. While this objection is a powerful threat to the foundation of AI governance, it tends to take normative interventions for granted. This can give the impression that the discussion of AI governance is nothing more than political rhetoric based mainly on individual moral views. This paper examines the truths and lies of the so-called “the problem of bias mirror” that AI decision-making reflects human bias like a mirror. AI sometimes reflects human bias as it is and sometimes creates additional harm (emergent harm). Since the latter is real, AI governance can be justified to some extent. On the other hand, AI offers us enormous material benefits or reinforces human ethics. So, even if the problem of bias mirror is real, these positive effects should be considered altogether, and a balanced view on AI governance should be taken into account.