Why does technical success with artificial intelligence (AI) in supply chain management (SCM) not always translate into economic value for manufacturing firms? This study examines the question through a structured integrative review of 36 core publications from 2010 to 10 August 2026, selected through eight anchor reviews, citation tracking, and targeted searches including reinforcement learning. Reported outcomes are classified into six performance levels (T1-T6), separately from five evidence levels. Representative publication levels are distributed as follows: Level 1, two; Level 2, seven; Level 3, 24; Level 4, three (one of which is a review- synthesized range); and Level 5, none. Among 25 non-review studies, 19 receive 34 overlapping performance-level assignments, while six inform adoption or implementation conditions. Five publications report financial outcomes, comprising two model- based cost evaluations, two interview or case-based benefit reports, and one annualized projection. None within the selected set meets all four Level 5 requirements: objective financial effects, an adequate actual operating period, total cost of ownership, and comparison with an alternative. Publication-level classifications do not establish the evidence level of every claim. The analysis links five conversion gaps and six cause areas to six value-realization gates (G0-G5) for investment, operational integration, validation, scaling, and continuation or termination. A hypothetical predictive-maintenance scenario specifies responsible departments, data, review documents, and realistic constraints. Gate-specific applications to generative, agentic, and reinforcement-learning AI are also presented. The framework is a literature-based managerial proposal, not a field-validated intervention; the results do not estimate industry-wide effects or failure rates.

1. 서 론
2. 연구 배경 및 연구방법
    2.1 경제적 가치와 가치실현 관점
    2.2 AI-SCM 선행연구와 연구 공백
    2.3 연구 설계와 자료선정
    2.4 코딩 및 분석방법
3. 분석 결과: AI-SCM의 가치실현과 성과격차
    3.1 SCM 기능별 성과와 근거의 범위
    3.2 핵심결과
4. 제조기업의 AI-SCM 가치실현 관리 정책
    4.1 단계별 가치실현 게이트
    4.2 총소유비용과 성과관리
    4.3 사람․거버넌스와 불확실성 관리
    4.4 제조기업의 AI-SCM 가치실현 실행원칙
    4.5 가상 적용 시나리오와 현실적 제약
5. 결론 및 시사점
    5.1 연구결과의 요약
    5.2 시사점
    5.3 연구의 한계 및 향후 연구
Acknowledgements
References
부록 A. 36편 핵심 근거문헌의 분석 지도
부록 B. 문헌선정․코딩 및 근거 추적자료

• Jonghyuk Joo(Department of Bio Industrial Engineering, Cheongju University) | 주종혁 (청주대학교 바이오산업공학 전공) Corresponding author