人工智能算力互联互通:生成逻辑、制度梗阻及其分层治理.
This article is based on the era context of the "Digital China" strategy and the development of new quality productive forces. It focuses on the institutional challenges emerging in the implementation of China's national strategic initiative of computing power interconnection, such as the inefficien...
| Publicado en: | Journal of Xi'an Jiaotong University (Social Sciences) Vol. 46; no. 3; pp. 150 - 163 |
|---|---|
| Autor principal: | |
| Formato: | Artículo |
| Publicado: |
Journal of Xi'an Jiaotong University (Social Sciences) Editorial Office
2026
|
| Materias: | |
| Acceso en línea: | Ver este registro en EBSCOhost |
| Sumario: | This article is based on the era context of the "Digital China" strategy and the development of new quality productive forces. It focuses on the institutional challenges emerging in the implementation of China's national strategic initiative of computing power interconnection, such as the inefficiency of computing resource allocation, the ambiguity of stakeholder responsibilities, and the lag in regulatory mechanisms, and conducts systematic institutional research. Against the backdrop of artificial intelligence driving the reconstruction of production factors, the quasi-public nature, high spillover, and cross-domain characteristics of computing power have led to the functional inadequacy of traditional administrative regulation and market allocation models. Therefore, this article adopts an interdisciplinary analytical approach, integrating administrative law theory, the economic theory of public goods. and governance theory, to construct a five-dimensional governance framework covering supply system, stakeholder responsibility. behavioral norms, resource allocation and full-process supervision. The core innovation of this article lies in four breakthroughs. Firstly, it provides a layered governance framework for computing power interconnection, offering theoretical support and institutional references for the future advancement of China's national computing power interconnection initiative. Secondly, it proposes the quasi-public goods positioning of computing power resources. thereby expanding the theoretical boundary of benefit administration in digital infrastructure governance. Thirdly, it designs a dynamic allocation mechanism composed of "national priority projects inclusive quotas-market bidding". aiming to achieve a balanced coordination among security, efficiency, and equity. Fourthly, it reconstructs a full-process regulatory system, advocating the establishment of a governance mechanism of "ex ante classified licensing-interim dynamic supervision-ex post exit assurance". The study further explains the institutional logic of computing power interconnection from three dimensions; building a unified national market, fulfilling the government's duty of benefit administration, and enhancing China's comparative advantage in the global competition for artificial intelligence. It identifies that in practice, computing power interconnection faces problems of regional imbalance. blurred accountability, lack of behavioral norms. inefficient distribution, and regulatory vacuums. To this end, it proposes a layered governance model covering the supply, subject, behavioral, allocation, and regulatory levels. At the macro-level, it emphasizes regional coordination and planning integration; at the meso-level. it clarifies the division of authority and responsibility among governments, enterprises, and industry organizations; and at the micro-level, it improves trading rules and behavioral standards, while institutionalizing the governance of computing power through differentiated allocation mechanisms and a full-process regulatory framework. The conclusion of this article indicates that the essence of computing power interconnection lies in the institutional reconfiguration of digital production factors. Its governance should not only enhance technical interoperability but also improve the institutional supply of regulatory rules through a hierarchical and multi-layered governance model. The policy implications are mainly reflected in constructing a differentiated allocation mechanism of "national priority-inclusive quota market bidding" to optimize the hierarchical scheduling of computing resources, and establishing a full-process regulatory framework of "ex ante classification-dynamic interim supervision-ex post exit assurance" to improve the regulatory system. The academic contribution lies in bridging the gap between legal theory and technological governance by advancing the dual paradigms of "computing power quasi-public-goodsization" and "layered governance". It further constructs a "layered governance framework for computing power" and an "institutional interoperability mechanism". providing a Chinese theoretical contribution and governance model for computing power legislation and global artificial intelligence infrastructure governance. 算力作为人工智能发展的核心引擎, 是当前最为重要的新质生产力之一。为促进算力资源的优化配置, 国家将算力互联互通作为阶段性目标。算力互联互通不仅与中国建设全国统一大市场的现实需求相契合, 也是政府履行给付行政义务的必要举措, 更是中国在全球人工智能发展竞争中取得比较优势的必然要求。算力互联互通既要求技术层面的设施互联互通, 又需要在制度层面予以有效的规则供给。针对推进算力互联互通过程中面临的区域分布配置失衡、主体职责定位不清、行为规范供给不充分、资源分配机制效果不佳、运营监管机制不完善等制度困境, 未来可通过细化区域统筹以强化算力供给, 厘清职责以提升主体协同, 完善规则以规范市场行为, 采用差异化机制以优化资源分配, 并实施全过程监管以保障治理效能, 从而构建覆盖供给层、主体层、行为层、分配层和监管层的算力互联互通分层治理框架, 为算力资源的高效配置提供具有可操作性的解决方案, 加快推进人工智能算力互联互通的有序实现。 |
|---|