An Analysis of the Multilayered Structure of Global AI Ethics Governance

다국어 초록 (Multilingual Abstract)

Purpose: The governance of global AI ethics is not about declaring the legitimacy of AI ethics per se, but rather analyzing the multi-layered nature of governance, where ethical principles are translated into actual norms, poli-cies, standards, procurement, auditing, and accountability systems. Therefore, the goal is to uncover the following: First, it clarifies the layers of global AI ethics governance and the regulatory instruments used at each layer. Sec-ond, it clarifies where coherence and conflict arise between layers, and what mechanisms mediate them. Third, it clarifies how the path from soft regulation, ethics, to quasi-norms or quasi-enforcement, is formed. This leads to proposals for the governance of AI ethics.
Method: This study first utilizes a literature review method. It first explores documents that present basic theo-ries related to governance theory and AI ethics policy practice. Next, it examines policy-related documents. Fur-thermore, some of the content encompasses multi-layered documents containing ethical standards, reports from Big Tech-focused companies, and audit frameworks. Next, it utilizes a comparative analysis method. The previous-ly discussed documents are compared by defining categories such as principles and values, obligations or re-quirements, sanctions and auditing as enforcement, and scope of application. Finally, it utilizes a developmental research method. This developmental research develops and presents a governance mapping structure.
Results: A structural analysis of global AI ethics governance at the international level concretizes the produc-tion of principles and competition for legitimacy. At the regional and supranational levels, it unfolds as a strengthening of norms through market integration. At the national level, it manifests as a combination of admin-istrative and supervisory systems and industrial policies. At the industry and sector levels, it converges into a risk-based, detailed framework. At the organizational level, internal governance is institutionalized around ethics committees. At the technological and system level, it is concretized as the codification of ethics, particularly in the context of data governance.
Conclusion: At each level, conflicts of value, jurisdiction, responsibility, and technology can arise. Furthermore, competition can arise between regulators in the establishment of norms, the market power of large platforms and cloud providers, standards-setting entities, and auditing power related to the evaluation criteria for market com-pliance. This will require the establishment of meta-principles that analyze and connect implementation require-ments, interoperability strategies among entities necessary to resolve conflicts at each level, and the internaliza-tion of accountability and redundancy mechanisms. Based on this, it is suggested that for specific countries or actors to secure initiative, they need to participate in strategic standardization, establish procurement standards, and establish industry-specific guidelines to become both adopters and producers of global norms.