作者/Author(s): Simon Vaughn 

網站來源/Source: Foreign Policy Research Institute 

日期/Date: 09/10/2025 

關鍵字/Keywords: 科技、中國、AI 


摘要：
雖然中國已制定了全面的 AI 戰略，但其中仍存在侷限性。此外，中國的AI政策雖對美國構成風險，但美國可憑藉自身優勢，維持科技領域的領先地位。
  • 中國的侷限： 
  1. 最新的 AI 模型需要最先進的運算硬體支援。中國雖已研發出本土替代方案，有效規避美國對高階 AI 晶片的限制，但這些方案效率低落、缺乏可擴展性，到頭來反而拖累了頂尖模型的研發進度。 
  2. 另一大挑戰則是預算限制。雖然中國中央與地方政府可以大力補貼，但也必須削減其他支出或重新分配資金來抵銷成本。若補貼無法立即產生經濟效益，這些投資恐血本無歸。 
  3. 中國的 AI 發展亦可能與自身國家安全優先事項相互衝突。開發者必須依循中央法規與管制來開發模型，這或將減緩 AI 創新的步伐。 
  • 若美國未能認清並提出應對中國人工智慧普及戰略的做法，將面臨多重戰略威脅： 
  1. 若美國未能將人工智慧融入其教育與研究體系，將在人工智慧的長期發展中落後中國。 
  2. 聯邦與各州間缺乏一致的 AI 政策，恐導致資源浪費與監管不一，反而拖慢 AI 在國內經濟普及的速度。 
  3. 美國同時也面臨人工智慧供應鏈的脆弱性。雖然美國企業負責主導設計AI晶片，但在製造環節卻高度依賴其他國家。 
  4. 中國的「軍民融合」計畫，讓解放軍能迅速將民用科技發展引進軍事領域；而美國軍民部門較為隔閡，很難複製這種模式。 
  5. 美國應該關注中國在塑造全球AI規範與標準方面越來越強大的影響力，這些規範可能反映其國家價值觀。 
  • 美國應聚焦自身優勢，並結合中國普及模式中的成功經驗，持續領導「負責任的 AI 部署」 
  1. 美國可提供針對性激勵措施，降低企業引進人工智慧的初始門檻，讓地方實驗與民建創新在聯邦政府指導下得以推進。 
  2. 美國擁有私部門主導的強大科技生態系統，但必須推動「普及政策」，縮短發明與普及之間的差距。聯邦與州政府可透過以下方式加速人工智慧普及：優先在基礎公共產品領域採用 AI、設立私部門的仿效KPI與目標，並提供明確指引與激勵措施，來規範並加速 AI 的普及與負責任使用。 
Summary: 
Despite a comprehensive strategy, China's AI strategy had its limitations. Furthermore, China's AI policies pose risks to the US, but it can leverage its advantages to maintain its lead position in the technology sector. 
  • China's limitations: 
  1. The newest AI models require access to the most advanced computing hardware. China has developed indigenous alternatives to circumvent US restrictions on the most advanced AI chips. However, these alternatives are inefficient and lack scalability, thereby slowing the progress in frontier model development. 
  2. Budgetary constraints also pose a challenge. The central and provincial governments may subsidize as they like, but they need to reduce spending on other matters or reallocate funds to offset the costs. Furthermore, these investments will go bust if the subsidies do not generate immediate economic returns. 
  3. China's AI development could contradict its national security priorities. Consequently, developers must create AI models that adhere to the central government's regulations, which may slow down AI innovation. 
  • The US faces several strategic threats if it does not recognize and respond to China's AI diffusion strategy: 
  1. The US will lag behind China in the long-term development of AI if it does not integrate AI into its education and research systems. 
  2. The incoherence of US AI policies between the state and central governments will lead to wasted resources and regulatory inconsistencies, subsequently slowing the diffusion of AI into its domestic economy. 
  3. The US also faces AI supply chain vulnerabilities. Although US companies dominate the design of AI chips, much of the manufacturing depends on foreign sources. 
  4. China's civil-military fusion program enables the PLA to integrate advancements in civilian technologies into its military, which is more challenging for the US to replicate due to divisions between its civilian and military sectors. 
  5. The US should take note of China's increasing influence in shaping global AI norms and standards, which could reflect its national values. 
  • The US should focus on its advantages and integrate them with successful lessons from the Chinese diffusion model to remain a leader in the responsible deployment of AI. 
  1. The US could provide targeted incentives to lower the bar for early-stage adoption of AI, allowing local experimentation and private innovation under federal guidance. 
  2. The US had a strong private-sector-led technology ecosystem, but it must promote adoption policies to reduce the gap between invention and diffusion. The federal and state governments could accelerate the diffusion of AI by prioritizing AI solutions in essential public goods, setting targets for private-sector replication, and providing clear guidance and incentives to regulate the responsible use of AI. 
 
回上層