Scholarlink

<p>Generative AI systems in Over-The-Top (OTT) platforms have transformed metadata production, yet they introduce critical governance challenges: processes exhibiting opacity, cultural bias, and accountability gaps that undermine equitable information access. To address these governance gaps, this paper introduces the Co-Constructive Metadata Governance Framework (CMGF), a human-centric model organized around four interdependent commitments: Adaptive Transparency, human-AI complementarity, cultural context preservation, and distributed accountability. The framework was developed through a sequential mixed-methods approach, moving from expert interviews to iterative design and empirical validation. This process explicitly integrates human values into the system design, rather than treating them as secondary considerations. Findings from the validation study (<i>N</i> = 125) indicate a strong preference for hybrid human–AI verification arrangements, with a majority of participants favoring collaborative approaches over fully automated systems. The CMGF provides implementable mechanisms—confidence scoring, Cultural Context Alert System(CCAS), and provenance tracking—that operationalize responsible AI principles, enabling platforms to leverage generative AI efficiency while preserving transparency, cultural sensitivity, and accountability essential for equitable information access in global digital ecosystems.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A Co-Constructive Metadata Governance Framework for Human-AI Collaboration in Generative OTT Ecosystems

  • 期刊论文

  • 发表日期：

    2026年7月6日

查看全文
  • Byungchul Cho,
  • Sunghee Ahn

摘要

Generative AI systems in Over-The-Top (OTT) platforms have transformed metadata production, yet they introduce critical governance challenges: processes exhibiting opacity, cultural bias, and accountability gaps that undermine equitable information access. To address these governance gaps, this paper introduces the Co-Constructive Metadata Governance Framework (CMGF), a human-centric model organized around four interdependent commitments: Adaptive Transparency, human-AI complementarity, cultural context preservation, and distributed accountability. The framework was developed through a sequential mixed-methods approach, moving from expert interviews to iterative design and empirical validation. This process explicitly integrates human values into the system design, rather than treating them as secondary considerations. Findings from the validation study (N = 125) indicate a strong preference for hybrid human–AI verification arrangements, with a majority of participants favoring collaborative approaches over fully automated systems. The CMGF provides implementable mechanisms—confidence scoring, Cultural Context Alert System(CCAS), and provenance tracking—that operationalize responsible AI principles, enabling platforms to leverage generative AI efficiency while preserving transparency, cultural sensitivity, and accountability essential for equitable information access in global digital ecosystems.