Earticle

초록

영어

This paper critically examines the phenomenon of the ‘outsourcing of ethics’ brought about by the expansion of automated decision-making systems (ADM), one of the core technologies of artificial intelligence, and explores the practical possibility of a Zen Buddhist AI ethics through the doctrine of the Three Causes of Buddha-nature(三因佛性) in the Seonjong yeongga jip(禪宗永嘉集), a classical text of the Zen Buddhist tradition. In contemporary society, ADM has become deeply involved in major domains of human judgment, including employment, criminal justice, and welfare administration. However, the automation of judgment is not merely a technological advancement that enhances efficiency and objectivity. It also carries the risk of reproducing prejudices accumulated throughout human history under the name of data, concealing responsibility for judgment behind technical procedures, and reducing human beings to predictable units of information. Based on this critical perspective, this paper examines three cases: Amazon’s AI recruiting system, the COMPAS recidivism risk assessment algorithm, and Australia’s Robodebt scheme. The case of Amazon’s AI recruiting system shows that when discriminatory practices of the past are embedded in data, algorithms may reproduce them as if they were objective evaluation criteria. The COMPAS case reveals that even when the same computational standard is applied, differences in social conditions and base rates among groups may impose a disproportionate burden of error on particular groups. The Robodebt case further shows that the automation of welfare administration can transform vulnerable citizens from subjects of care into objects of suspicion, surveillance, and proof. These three cases suggest that the ethical problem of ADM is not simply a matter of technical error, but a structural crisis that arises when the principles of fairness, accountability, transparency, and human oversight are not sufficiently realized within concrete institutional contexts. To critically clarify these limitations, this paper focuses on the eighth question, ‘Clarifying the Expository Means and the Ultimate Meaning’(簡其詮旨), in the ‘Song of Upekkh (優畢叉頌)’ section of the Seonjong yeongga jip. The relationship among exposition(詮), purport(旨), contemplative practice(觀) and fundamental principle(宗) presented in this passage is connected in the commentarial tradition with the structure of the Three Causes of Buddha-nature: the direct cause of Buddha-nature(正因佛性), the complete cause of Buddha-nature(了因佛性), and the auxiliary cause of Buddha-nature(緣因佛性). This paper reinterprets this structure from the perspective of AI ethics. The direct cause of Buddha-nature reminds us that human beings possess a fundamental possibility that can never be reduced to data or predictive values. The revealing cause of Buddha-nature calls for an awakened wisdom that does not uncritically accept algorithmic outputs, but instead reflects on the data, standards, and conditions of discrimination that make such judgments possible. The conditional cause of Buddha-nature shows that this awakening must not remain an abstract principle, but must be realized within concrete institutions and practices through conditions such as rights remedies, explainability, accountability structures, and human oversight. In conclusion, the Zen Buddhist AI ethics proposed in this paper does not reject technology itself or deny the possibility of automation. Rather, while acknowledging the usefulness of algorithms, it emphasizes that human judgment can never be entirely delegated. The aim of AI ethics cannot remain limited to the engineering goal of implementing an error-free algorithm. It must also move toward establishing institutional and ethical conditions that enable human beings not to be exhausted as mere outputs of data, but to exercise responsible judgment and awakened reflection for themselves. In this respect, the doctrine of the Three Causes of Buddha-nature in the Seonjong yeongga jip provides a philosophical resource for deepening the discussion of fairness, accountability, and transparency in the age of ADM into questions concerning the possibility, awakening, and practical conditions of human existence.

한국어

인공지능 핵심 기술 가운데 하나인 자동화 의사결정 시스템(ADM)은 채용, 형사사법, 복지 행정 등 인간 삶의 주요한 판단 영역에 깊숙이 개입하고 있다. 그러나 판단의 자동화는 단순히 효율성과 객관성을 높이는 기술적 진보에 그치지 않고 인류 역사를 거쳐 축적되어 온 편견을 데이터의 이름으로 재생산하고, 판단의 책임을 기술적 절차 뒤로 은폐하며 인간을 예측 가능한 정보 단위로 환원할 위험을 동반한다. 이러한 문제의식을 바탕으로 아마존 AI 채용 시스템, COMPAS 재범 위험평가 알고리즘, 호주의 로보뎁(Robodebt) 사례를 검토한다. 아마존 AI 채용 사례는 과거의 차별적 관행이 데이터 속에 축적될 때 알고리즘이 이를 객관적 평가 기준으로 재생산할 수 있음을 보여준다. 이러한 한계를 비판적으로 해명하기 위해 본 논문은 『선종영가집(禪宗永嘉集)』 「우필차송(優畢叉頌)」의 제8문 ‘간기전지(簡其詮旨)’에 주목한다. 이 대목에서 제시되는 언전(詮)과 취지(旨), 관행(觀)과 근본(宗)의 관계는 주석 전통 속에서 정인불성(正因佛性), 요인불성(了因佛性), 연인불성(緣因佛性)의 구조와 연결된다. 본 논문은 이를 AI 윤리의 관점에서 재해석하였고 결론적으로 禪적 AI 윤리는 알고리즘의 유용성을 인정하되, 인간 판단이 결코 전적으로 위임될 수 없다는 점을 강조한다. AI 윤리의 지향점은 오류 없는 알고리즘의 구현이라는 공학적 목표에만 머무를 수 없다. 이러한 맥락에 있어 『禪宗永嘉集』의 삼인불성론은 ADM 시대의 공정성·책임성·투명성 논의를 인간 존재의 가능성, 자각, 실천적 조건의 문제로 심화시키는 사상적 자원을 제공할 수 있다.

목차

국문 초록
Ⅰ. 서론
Ⅱ. ADM과 알고리즘 윤리의 구조
1. 사례 1-AI 채용 시스템과 젠더 편향
2. 사례 2-재범 위험평가와 예측적 정의의 문제
3. 사례 3-복지·공공 행정 ADM과 취약계층 배제의 구조화
Ⅲ. ADM에 대한 禪적 윤리 비판
1. 三因佛性의 중도적 보편성
2. 『禪宗永嘉集』 「優畢叉頌」의 三因佛性
3. ADM에 대한 禪적 윤리 비판
Ⅳ. 결론
참고 문헌
Abstract

키워드

저자

• 문유정 [ Mun, Yu-Jeong | 동국대학교 불교학과 강사 ]

참고문헌

발행기관

간행물

표지보기 관심저널 등록

• 간행물명

  禪學(선학) [Journal of Korean Seon Studies]

• 간기

  연3회

• pISSN

  1598-0588

• 수록기간

  2000~2026

• 등재여부

  KCI 등재

• 십진분류

  KDC 228 DDC 294