초대형 국책 AI 프로젝트의 효과적인 통합관리방법론: DAINOS 프레임워크 적용을 중심으로

Abstract

This study introduces the DAINOS Framework as an integrated management approach for hyper-scale AI projects in manufacturing, characterized by technical complexity and multi-organizational collaboration. As open innovation becomes more common, large-scale AI initiatives increasingly involve diverse stakeholders such as data providers, AI developers, infrastructure suppliers, and users. Traditional Work Breakdown Structures (WBS) often fail to clarify roles, facilitate functional communication, or coordinate collaboration effectively. To address these limitations, DAINOS is proposed as a role-based meta-framework that complements WBS. It consists of six domains—Data, AI, Infrastructure, Network, Organizing, and Service—and provides functional context to tasks through auxiliary labeling. This enables clear role alignment and structured collaboration. Grounded in systems engineering, complexity theory, and collaborative network theory, DAINOS was applied to a large-scale AI service development project involving 15 institutions. Results showed it functioned as a common language among stakeholders, reducing confusion and enhancing project governance. In qualitative assessments, stakeholders reported clearer role recognition, reduced inter-organizational friction, and more effective communication; for example, 15 participating institutions achieved streamlined decision-making processes and observed measurable improvements in collaboration efficiency. The study contributes both theoretical insights and practical tools for managing complex socio-technical systems. DAINOS also shows promise as a generalizable framework applicable to sectors such as aerospace, defense, and healthcare. Future work will focus on quantitative evaluations, cross-industry comparisons, and developing a maturity model for framework evolution. In summary, these research directions aim to further develop DAINOS into a robust and generalizable governance framework with practical applicability and scalability, thereby enhancing project governance and collaboration in hyper-scale AI projects and similar multi-organizational R&D initiatives, and ultimately contributing to diverse industries.