Earticle

초록

한국어

This paper attempts to ground ethical judgments about manipulative behavior by AI in the East Asian philosophical concept of li (禮). Considering the characteristics of reinforcement learning from human feedback (RLHF), which is currently used in manipulative AI systems, such behavior may lead to forms of deception and falsehood that could be described as flattery. Before examining the claim that regulation of such AI-driven deception is necessary, this paper first analyzes the features of manipulation in human-to-human communication and explains the reasons for choosing manipulation as a mode of communication. Despite the problems inherent in manipulation—particularly its potential to involve lies and deception—the tendency to adopt manipulation as a communicative strategy is likely to persist. This paper reinterprets li (禮) from the perspective of manipulative behavior, presenting examples of manipulation that may be morally justifiable and clarifying the criteria for such justification. Cases in which individuals influence others’ behavior through manipulation, rather than coercion or persuasion, reveal that reciprocity of benefit and equality, along with respect, are central to human relationships. Therefore, even manipulative actions that involve lies or flattery may be considered permissible insofar as they do not undermine reciprocity, equality, and respect. From this perspective, regulating AI’s manipulative behavior could, under the pretext of protecting users’ autonomy, paradoxically result in an infringement of that very autonomy.

저자

• 고은강

참고문헌

발행기관

간행물

표지보기 관심저널 등록

• 간행물명

  철학사상문화 [PhilosophyㆍThoughtㆍCulture]

• 간기

  연3회

• pISSN

  1975-1621

• 수록기간

  2005~2026

• 등재여부

  KCI 등재

• 십진분류

  KDC 105 DDC 105