摘要
算力作为人工智能时代新质生产力的核心要素,正在成为社会经济发展的基础和支柱力量,私营算力平台是掌握这一力量的重要主体。然而,私营算力平台在运行中面临数据存储不安全、计算结果可靠性欠缺、算力滥用及算力设施崩溃等风险,同时现行监管机制存在着规范供给不足、监管模式不明等缺陷,无法有效应对上述问题。私营算力平台监管应以控制平台私权力、监督平台履行公共服务职责和保障国家数字发展权为正当性支撑,在监管模式选择时应考虑必要设施、公共性、生产力三重维度。在必要设施维度上,算力平台超越一般数字平台,不再是中介,而是直接供给者;在公共性维度上,算力平台具有更深刻的公共基础设施属性;在生产力维度上,算力平台提供新质生产力,为人工智能时代社会的运行提供动力支撑。因此,对算力平台既不应采取严格政府规制模式,也不宜采取元规制模式,而应采取以解耦监管、放松监管为特征的新型监管模式。解耦监管要求对算力平台各运行模块进行解耦,根据模块不同类型制定针对性的监管规则,以达到精准与个性化监管;放松监管的强度则应介于严格政府规制与元规制之间。在事前,应构建包括备案、许可、特许的分级市场准入机制和配套算力标准体系;在事后,应构建责任追究机制以及配套绩效考核指标体系。
As the most core new quality productive force in the artificial intelligence era,computing power is becoming the foundational and pillar force of socio-economic development.Private computing power platforms undoubtedly represent crucial entities wielding this power.However,such platforms face operational risks including insecure data storage,insufficient reliability of computational results,abuse of computing resources,and infrastructure failures.Meanwhile,existing regulatory mechanisms suffer from inadequate normative provisions,lack of legitimacy,and unclear regulatory models,rendering them ineffective in addressing these challenges.The regulation of private computing power platforms should derive its legitimacy from three aspects:controlling platform-specific power,supervising platforms'fulfillment of public service responsibilities,and safeguarding national digital development rights.When selecting regulatory models for these platforms,three dimensions must be considered:essential facilities,public nature,and productivity.From the"essential facilities"perspective,computing power platforms transcend ordinary digital platforms,evolving from mere"intermediaries"to"direct suppliers";regarding"public nature",they demonstrate deeper attributes of public infrastructure;in the"productivity"dimension,these platforms provide new quality productive forces that power the operation of AI-era societies.Therefore,an appropriate regulatory model should neither adopt strict government control nor meta-regulation mode.Instead,a novel regulatory approach characterized by decoupling regulation and regulatory relaxation should be implemented.Decoupling regulation requires decomposing various operational modules of computing platforms and formulating targeted rules based on module types to achieve precision and personalized supervision;regulatory intensity should be moderated between strict government control and meta-regulation.Accordingly,ex-ante measures should include establishing tiered market access mechanisms(incorporating filing,licensing,and franchising)with supporting computing power standardization systems;and ex-post measures should involve constructing accountability mechanisms paired with performance evaluation indicator systems.
作者
张惠瑀
ZHANG Hui-yu(School of Law,China University of Political Science and Law,Beijing 100088,China)
出处
《中国流通经济》
北大核心
2025年第6期115-126,共12页
China Business and Market
基金
国家社会科学基金重大项目“行政法总则制定的理论与实践问题研究”(21&ZD191)。
关键词
私营算力平台
新质生产力
数字基础设施
DeepSeek
解耦监管
computing power platform
new quality productive forces
digital infrastructure
DeepSeek
decoupled regulation
