탈인본주의적 AI 윤리란 무엇인가?:‘윤리적 AI’를 중심으로

박휴용 한국컴퓨터교육학회 2022 컴퓨터교육학회 논문지 Vol.25 No.6

This paper explores the possibilities and rationales of post-humanistic ethics as the foundation of AI ethics, in the trends of the role of human being has been superseded by that of AI. This paper regards posthumanism as an alternative ethico-philosophical basis for AI ethics, and pays attention to the term, ‘ethical AI’, which represents AI as the ethical subject. Two main research questions were conceived for this paper’s discussion: first, can AI replace the role of human-being as an ethical subject by performing higher-level autonomous judgment and action?; and second, can posthuman AI ethics overcome the limitations and dilemmas of previous AI ethics by suggesting alternative ethical provisions for the future? To address these questions, this paper discussed relevant terms of AI ethics, and compared the differences of humanist and post-humanistic AI ethics by highlighting the necessities of posthuman ethical perspectives, and, finally, critically reviewed the limitations and challenges of current AI ethical principles. This paper’s argument may entice the discussions of academia for the alternative directions of AI ethics in the rapidly changing future society. 본고는 인공지능이 점점 인간의 역할의 대신하여 미래 사회의 핵심적 구성원이 되어감에 따라 인공지능 윤리의 이론적 기초로서 탈인본주의적 윤리관의 필요성을 강조하고자 하였다. 이를 위해 본고는 포스트휴머니즘의 철학적 세계관을 기초로 AI를 인간 수준의 윤리적 주체로 상정하는 ‘윤리적 AI’(ethical AI)란 개념이 미래 인공지능 윤리의 방향성에 합당한지에 대해 논의하였다. 이를 위한 본고의 연구문제는 첫째, 인간의 역할을 대신하는 수준의 고차원적 인공지능은 자율적 판단과 행위를 수행함에 따라 독자적인 윤리적 주체가 될수 있는가와 둘째, 포스트휴머니즘에 기반한 AI 윤리가 기존의 AI 윤리 원칙들의 한계성과 딜레마를 극복하고 대안적인 인공지능 윤리로서 역할을 할 수 있는가이다. 이 문제를 다루기 위해 본고는 AI 윤리 관련 개념들-AI 윤리, AI-보조 윤리, 그리고 윤리적 AI-에 대해 설명한 다음, 인본주의와 탈인본주의적 AI 윤리의 차이점을 대조하면서, 인본주의적 윤리관의 대안으로서 포스트휴머니즘에 기반한 AI 윤리의 가능성에 대해 논의하였고, 마지막으로 현재의 AI 윤리 원칙들의 한계성, 탈인본주의적 윤리관을 기반으로 한 윤리적 AI에게 주어진 도전들, 그리고 ‘윤리적 AI’의 확장을 위한 고려사항이 무엇인지에 대해 논의하였다. 본고는 인공지능이 점차 윤리적 주체로서의 역할을 수행해가는 미래 사회를 대비한 인공지능 윤리의 방향성에 대한 학계의 논의에 일조할 수 있을 것이다.

게이미피케이션 기반 인공지능 윤리교육 프로그램이 초등학생들의 인공지능 윤리 의식에 미치는 영향

원호준,이철현 경인교육대학교 교육연구원 2023 교육논총 Vol.43 No.4

This study investigated the effects of a gamification-based education program centered on Artificial Intelligence (AI) ethical standards on elementary school students’ ethical awareness of AI. Therefore, a 16-session AI ethics educational program was developed based on the ADDIE model according to previous studies on AI ethics education and “human-centered AI ethical standards.” To verify the program’s effectiveness, the program extracted from the “AI ethics (Elementary School) Learning with AI Principles” (Gyeonggi Provincial Office of Education, 2022) was applied to compare between groups and used a student ethics consciousness test tool we developed. The study results obtained through the application and analysis of this program are as follows. We found statistically significant differences in all areas of interest and the need for AI ethics and ethics education. The results of the pre-and post-tests for each AI ethics element showed statistically significant changes in all areas, except for the “responsibility” element. The AI ethics education program developed in this study is expected to contribute to the formation of a proper AI ethics consciousness among elementary school students.

토픽 모델링을 활용한 국내 AI 윤리 연구 동향 및 주요 이슈 분석

박보경,윤가영 한국교육공학회 2025 교육공학연구 Vol.41 No.2

This study conducted a trend analysis, keyword analysis, and Latent Dirichlet Allocation (LDA)-based topic modeling on 361 academic papers on artificial intelligence (AI) ethics published in Korean journals from 2023 to 2024, where generative AI technologies, such as ChatGPT, were widely adopted. The findings are as follows. First, research on AI ethics surged in 2023, coinciding with the emergence of generative AI. Research was actively conducted in the social sciences and humanities, encompassing education and law. Second, domain analysis based on the literature themes found that a significant portion of the research was conducted on “AI ethics education” and “laws, institutions, and norms,” with particular emphasis on the necessity and practical strategies for AI ethics education, especially in elementary education. Third, keyword analysis identified core ethical and social values, such as “education,” “responsibility,” “morality,” and “literacy,” as central concepts in AI ethics research. Fourth, topic modeling by research domain indicated that discussions on “ethics of AI” focused on logical reflection and responsibility ethics; “ethics in the use of AI” emphasized reestablishing values in social, public, and industrial sectors; “ethics in educational applications” addressed teachers' and learners' perceptions and educational practices; and “educational ethics” examined curriculum integration and instructional strategies. The study's results provide a comprehensive understanding of AI ethics discourses, suggesting implications for the future development of ethics education and policy initiatives.

규범적 AI윤리 가이드라인과 초등 AI윤리교육 적용 방안 모색

박형빈 ( Park Hyoung Bin ) 서울교육대학교 초등교육연구원 2024 한국초등교육 Vol.35 No.2

The advent of artificial intelligence (AI) technology is driving revolutionary changes across various sectors of contemporary society, encompassing industry and education, bringing considerable convenience to our lives. Nevertheless, as AI progresses, various ethical concerns emerge, such as invasions of privacy, bias and discrimination, ambiguity in responsibility, and military applications, heightening the necessity for societal regulation. Particularly, systems such as large language models frequently function as “black boxes” with internal mechanisms that are not transparent to users, resulting in challenges understanding and trusting AI outputs. To tackle these ethical concerns, we must establish a process that enables the trustworthiness and validation of AI outcomes. This approach should prioritize the strict adherence to ethical standards and protocols at every stage from development to use. The importance of systematic AI ethics education for primary school students, who will assume roles as AI designers, developers, policymakers, and users in the future is becoming more apparent, as teaching them the proper use and ethical responsibilities associated with AI technology is crucial. This study aims to identify educational measures to address ethical problems posed by AI and proposes incorporating normative AI ethics guidelines into primary AI ethics education. It investigates the principles covered by existing AI ethics guidelines and identify their limitations. It also attempts to answer the following questions: What are the functional foundations of human morality as revealed by neuroscience, and what considerations should be made during the ethical decision-making process of AI? Why are normative AI ethics guidelines necessary, and how can they be implemented in primary AI ethics education? Ultimately, the primary objective of AI ethics education should be to gain a comprehensive understanding of AI-related ethical issues and cultivate the ability to embrace the beneficial changes AI technology brings, while effectively managing potential risks and ensuring its ethical use. This study aims to cultivate the multifaceted skills required for elementary school students to form ethical judgments and make decisions related to AI.

초등학교 AI 윤리교육 프로그램의 개발 및 적용

정다이,박형빈 서울교육대학교 초등교육연구원 2023 한국초등교육 Vol.34 No.3

AI 개발은 그 어떤 변화보다 인간의 삶을 뒤흔드는 혁명적인 변화를 불러일으키고 있다. 데이터 처 리와 분석을 재편하는 머신러닝, 딥러닝, 인공신경망과 같은 AI 접근 방식은 의료, 운송, 생산라인 등 다양한 분야에서 자율 및 반자율 시스템으로 활용되고 있다. 그러나 AI의 행위와 관련된 윤리 문제는 사회 윤리의 일부로 여겨지고 있는 현실이다. AI가 ‘윤리적이어야 한다.’라는 동의에도 불구하고, ‘윤 리적 AI’를 정의하고 실현하기 위한 요소, 윤리적 요구사항, 기술 표준, 모범 사례 등에 대한 논쟁은 계속되고 있는 것이다. 특히 이때 AI는 교육 분야에서도 중요한 역할을 한다. 미래 AI의 설계자, 개발 자, 사용자, 평가자로서의 역할을 할 초등학생들에게 AI 윤리 교육으로부터 얻을 수 있는 역량이 AI와 함께 살아갈 미래의 사회 구성원으로서 필수 역량으로 자리 잡을 것이기 때문이다. 따라서 초등학생에게 AI가 야기할 윤리적 문제를 조정하고 해결할 수 있는 능력을 기르기 위한 AI 윤리 교육이 필요하다. 이를 위해 본 논문은 인공지능 윤리 관련 내용을 구성하기 위하여 2022 개정 교육과정 중 도덕과 교육과정의 내용체계를 인공지능 윤리교육의 관점에서 분석한다. 그 후에 AI 윤 리와 관련된 필수 내용을 ‘자율주행 자동차, 사이보그, 뇌, 교육’의 4가지 주제 중심으로 체계화하여 수업안을 제안하였다. 마지막으로 개발한 수업안을 실제 수업에 적용하여 효과를 분석한다. AI 윤리 교수학습 프로그램을 현장에 선제적으로 구안 및 적용한 결과 다음과 같은 결론 및 시사점 을 도출할 수 있었다. 첫째, 인공지능 윤리 프로그램의 도입은 인지, 정의, 행동적 영역 모두에서의 도 덕적 사고력 및 AI 리터러시 증진을 이끈다는 점을 확인하였다. 둘째, 주제 중심의 프로그램 적용 후 학생들의 인공지능 윤리 학습에 대한 흥미도가 증대되었다. 셋째, 프로그램 중 개별 학습의 제공한 것 이 AI 윤리 문제와 관련한 학생들의 입장을 명확하게 하는 것에 도움이 되었다. 마지막으로 2022 개 정 도덕과 교육과정에 맞추어 AI 윤리의 핵심 내용이 추출되어 정리되고 새로운 기능, 역량이 추가되 어 현장에 보급될 필요성이 있음을 확인할 수 있었다. Artificial intelligence (AI) development is leading to a revolutionary transformation in human life. AI approaches, such as machine learning, deep learning, and artificial neural networks, which restructure data processing and analysis, are utilized in various fields, such as healthcare, transportation, and production lines, as autonomous or semi-autonomous systems. However, ethical issues regarding AI behavior remain a significant societal concern. Despite the consensus that AI should be ethical, debates are ongoing about defining and realizing ethical AI, which encompasses elements of ethical AI, ethical requirements, technical standards, and best practices, among others. Particularly, AI plays a crucial role in the field of education. Elementary-school students, who will become designers, developers, users, and evaluators of future AI, must develop the capacity to address and solve ethical issues that emerge from AI use. Hence, AI ethics education for elementary students is necessary for cultivating the skills required to become essential participants of a future society that extensively uses AI. To this end, this study analyzes the content and structure of the revised 2022 curriculum of Republic of Korea from the perspective of AI ethics education. It then proposes a lesson plan by organizing essential AI ethics–related content that focuses on four themes, namely, autonomous driving cars, cyborgs, the brain, and education. Finally, it assesses the effectiveness of the developed lesson plan through implementation in real-world classrooms. The study draws the following conclusions and implications after the proactive implementation of the AI ethics education program in classrooms. First, introducing AI ethics programs enhanced moral reasoning and AI literacy in the cognitive, definitional, and behavioral aspects. Second, the interest of students in AI ethics education increased after the implementation of the theme-centered AI Ethics program. Third, allowing individual learning within the program helped elucidate the position of students on AI ethics issues. Finally, the study extracted and organized the core content of AI ethics in the revised 2022 curriculum and suggested the need for the dissemination of AI ethics in the field with the addition of new features and capabilities.

인공지능윤리교육의 국내 연구 동향 분석

김경주,조정원 (사)디지털산업정보학회 2023 디지털산업정보학회논문지 Vol.19 No.4

This study examined research trends in AI ethics education and attempted to suggest a direction for AI ethics education. As a result of the research, two studies were conducted in 2017. There are no studies in 2018 and 2019, and there are 6 studies in 2020. Since then, research has continued to increase, with 19 studies in 2021 and 18 studies in 2022. There were a total of 37 lead authors of the study. There were six lead authors who had published papers for more than two years, and two lead authors who had published papers for more than three years. In addition, to examine the details of AI ethics education, a total of 265 keywords that went through a refining process were divided into education-related, ethics-related, AI-related, and other-related. Although the necessity and importance of research on AI ethics education is expected to increase, there are not many researchers who continuously conduct research on AI ethics education. Accordingly, there is a need to find ways to continue research on AI ethics education. AI ethics education is being conducted under various names such as moral education, ethics education, liberal arts education, and AI education. Accordingly, research on AI ethics education at various levels and forms should be conducted, not just educational research on artificial intelligence ethics in terms of regular subjects.

AI 윤리 교육의 필요성에 대한 연구

변순용 서울교육대학교 초등교육연구원 2020 한국초등교육 Vol.31 No.3

The main purpose of this study is to propose the necessity and content of AI ethics educationbased on the analysis of recent domestic and international trends related to AI ethics. Reflectingchanges in UNESCO's AI ethics recommendations, the announcement of the AI national strategy, andthe entry into force of the Framework Act on Intelligent Informatization, we should try to establish acontent system for AI ethics education and integrate it into the moral and curriculum. Therefore, itis necessary to discuss how to accept the new contents of the practical ethics of AI ethics in thecurrent elementary and secondary ethics and education. In this paper, the need for the establishment of basic AI competencies, the need for AIconvergence education experts, and the need for AI ethics and citizenship education was presented. As an example of the content structure of AI ethics education, we analyzed the contents of AI ethicseducation developed by MIT and the education program of Harvard University, and proposed anexample of the content system of AI ethics education that can be done in Korea. In particular, an opportunity to think about the importance of AI ethics education centering onthe topic of 'If artificial intelligence is in charge of knowledge transfer, what should a teacher do?'in the course of training artificial intelligence ethics experts or teachers is necessary to provide. Inthe age of artificial intelligence, education innovation and paradigm shift will surely occur, and it isan important time to reveal that the center is not technology, but artificial intelligence ethics. 이 연구의 주된 목적은 AI 윤리와 관련된 최근 국내외적인 동향에 대한 분석을 근거로하여 AI 윤리교육의 필요성과 내용에 대한 제안을 하는 것이다. 유네스코의 AI 윤리 권고안, AI 국가 전략 발표, 지능정보화기본법의 발효 등의 변화를 반영하여, 우리는 AI 윤리교육의 내용체계를 수립하여 도덕과 교육과정에 통합시켜보려는 노력을 해야 한다. 그래서 현행 초, 중등 도덕과 교육이 AI 윤리라는 실천윤리의 새로운 내용을 수용하는 방안에 대한 논의가 시급하다. 이 연구에서는 세계 여러 나라에서 개발되고 있는 AI 관련 초·중등학교용 교재를 통해 인공지능 교육이 이미 학교교육의 하나로 다뤄지고 있음을 제시했다. AI 윤리 교육의 내용 구성의 예시로, MIT에서 개발한 AI 윤리 교육내용과 하버드 대학의 교육프로그램을 분석하고, 한국에서 실시할 수 있는 AI윤리교육의 내용체계의 예시를 제안하였다. 이러한 변화에 더해서 인공지능 융합 교육 전문가, 인공지능 시민성 교육 전문가가 필요함을 언급하고, 더욱이 인공지능 역기능에 대응하는 것, 인공지능윤리전문가의 필요성을 강조하였다. 특히, 인공지능윤리 전문가 양성, 혹은 교사 양성과정에서 ‘인공지능이 지식전달을 담당한다면, 교사는 무엇을 해야 하나?’라는 화두를 중심으로 ’인공지능윤리 교육‘의 중요성을 생각해볼 수 있는 계기를 제공하는 것이 필요하다. 인공지능 시대는 교육의 혁신, 패러다임의 전환이 분명히 일어날 것이며그 중심은 기술이 아니라 인공지능윤리임을 드러낼 수 있는 중요한 시점이다.

학교 교육과정에서 융복합적 AI윤리교육의 과제 - 교육대학원 AI융합교육전공과 ‘AI 윤리’ 수업 사례를 중심으로 -

송선영 ( Song Sun-young ) 한국윤리교육학회 2022 윤리교육연구 Vol.- No.63

이번 연구는 교육부의 학교 교사 대상 융합교육 정책 지원에 따른 38개 교육대학원 AI융합교육전공(재교육) 운영과 ‘AI 윤리’ 관련 선택과목 수업 개설 현황, 그리고 경상국립대학교 교육대학원 AI융합교육전공의 <인공지능과 윤리교육> 수업 사례를 통해 학교 교육과정에서 AI윤리교육의 과제를 검토하고자 한다. 2020년 2학기부터 전국 38개 교육대학원에 AI융합교육전공(재교육)이 개설되었고, 이 가운데 ‘AI 윤리’ 관련 교과목(선택과목)을 운영하는 곳은 15개교였다. 경상국립대학교 교육대학원 수업 운영 사례를 통해 드러난 융복합적 AI 윤리교육의 문제와 과제는 다음과 같다. 첫째, ‘인공지능 윤리’에 관한 정의가 인공지능의 활용 범위에만 국한되어 있다. 둘째, 학교 교육과정에서 AI 윤리 관련 수업 설계가 교사의 담당 과목에 따라 교육내용이 다르게 전개될 수 있다. 셋째, 교과마다 기술적 적용 수준이 서로 다르기 때문에, 교수학습 방법에서도 일관된 연계를 설정할 수 없다. 이와 더불어 도덕윤리과에서 융복합적 AI 윤리 수업을 전개하기 위한 주요 과제는 다음과 같다. 첫째, 융복합적 교과에 도덕윤리과도 제도적으로 포함되어야 한다. 둘째, 학교 교육과정과 수업을 위해 도덕윤리 교사도 AI융합교육전공 이수를 고려할 필요가 있다. 셋째, 2022 교육과정 개편 과정에서 AI 윤리 관련 주제가 포괄적으로 적용될 필요가 있다. This paper aims to explore the task of AI ethics education in the school curriculum by examining the major of AI Convergent Education in 38 graduate schools of education in the second semester of 2020. In 15 graduate schools of education, the subject of AI ethics has been operated. By the experience of AI ethics in the graduate school of Gyeongsang National University, I would point out some problems as follows. Firstly, there are arbitrary definitions of ‘AI ethics’ that have made many teachers confused in its class design. Secondly, there are different levels, aspects and gaps of AI ethics education by main subjects of teachers among elementary, middle, and high schools. Thirdly, there are no common or acceptable teaching and learning methods of AI ethics class by the absence of the generalized curriculum of convergent AI ethics. In addition, I would suggest some tasks to develop AI ethics classes in ethics education. Firstly, there are also moral subjects in the scope of convergent subjects, such as mathematics, science, computer and information subjects, connected with AI. Secondly, some moral teachers need to take the AI convergent education in graduate schools of education. Finally, some topics of convergent AI ethics should be dealt with in one of revised or renewed elements in the plan of 2022 National Curriculum.

AI 윤리 기준을 활용한 AI 윤리교육 방안 연구

이인재 ( Lee In Jae ) 한국초등도덕교육학회 2022 초등도덕교육 Vol.- No.76

While innovative advances in AI-based technologies are positively impacting economic growth, social development, and human well-being, they also pose a risk to humanity and society as well as designers, developers, users, and people of AI by causing low explainability, data bias and privacy, and a variety of ethical issues. Therefore, appropriate AI ethical guidelines and AI ethics education are being emphasized to counter the negative effects of these AI systems. This paper aims to explore ways of AI ethics education by utilizing values or principles implied by AI ethical guidelines. To this end, I comprehensively analyzed the AI ethical guidelines set forth by domestic and foreign governments, research institutions, and private companies, and then drew up implications for AI ethics education and propose methods for systematic AI ethics education by utilizing the relationship between ethics of AI and ethical AI, which are elements of AI ethics.

Suggestion of Building the AI Code of ETHICS through Deep Learning and Big Data Based AI

김현수 J-INSTITUTE 2021 Robotics & AI Ethics Vol.6 No.1

Purpose: Current social and technical issues related AI Ethics take many different forms. Therefore, efforts to make ethical guidelines to cope with these issues are actively being developed. However, most kinds of ethical guidelines present general ethical principles and take a deductive method of solving individual problems in ac-cordance. The purpose of this study is to propose creating ethical guidelines through an inductive method of deriving ethical principles based on ethical judgements on each individual AI-related cases. Method: At first, most representative cases of AI ethics-related guidelines would be investigated in domestic and international level, with collecting documents and literature review. After that, examine the commonalities and differences between cases as these basic data through comparative research methods. Accordingly, it would be revealed that each case is constituted by a deductive method. Finally, as an alternative to these methods, presenting the merits of establishing ethical principles related to AI through inductive cases and specific examples. Results: Most of the representative AI Code of Ethics that currently exist have the form of suggesting principles and then solving ethical problems by applying the principles to the actual events accordingly. This type of approach corresponds to the method of ethics which based on moral principles. However, complex and unpre-dictable problems are likely to arise when it comes to AI ethics. In order to solve these problems, it is necessary to extract the principles of the AI Code of Ethics by establishing and presenting ethical principles through researching and analyzing various individual events related to AI Ethics using deep learning and Big Data Based AI. Conclusion: The following effects can be achieved by using deep learning techniques and Big Data Based AI that contains Ethical Issues together with sound and desirable Moral Judgement on each case, to derive the principles of the AI code of ethics. First, it is possible to extract and secure Big Data as basic resource of presenting Ethical Directions on various ethical problems arising in connection with the development of AI technology. Second, since the Ethical Principles as AI Code of Ethics are established based on empirical data, the validity of the principles can be secured. On the other hand, the AI Code of Ethics derived through deep learning based on such Big Data is likely to result in multiple tyranny or errors of majority due to certain limitations. So evaluation, verification and correction by Human Ethics Experts are essential to prevent these kinds of fault.