Risk-oriented artificial intelligence regulation: progress, innovation and inspiration from the European Union perspective
Wang Chunxiao，Li Huaisheng
School of Criminal Justice, China University of Political Science and Law
Abstract： As artificial intelligence technology advances and market application risks evolve, artificial intelligence regulation has emerged and been independent of artificial intelligence governance, which has become the basic concept and emerging field of resolving artificial intelligence risks. The European Union has pioneered a new paradigm of risk-oriented AI regulation, and realized the leadership and action of EU AI regulation by means of international cooperation, ethical constraints and risk regulation framework. Considering the difference between artificial intelligence technology and social needs, the EU should optimize the main body, regulatory process and regulatory content of artificial intelligence to achieve the effective regulation of artificial intelligence risks. Faced with the complex risk situation of artificial intelligence, China can learn from the experience of the European Union, and formulate the risk-oriented legal policy and framework system of artificial intelligence based on local national conditions at the policy decision-making level. At the level of application and practice, an interactive and collaborative mechanism among judicial organs, market supervision organs and enterprises is constructed. At the industrial structure level, the market empowerment was used to promote the sinking of artificial intelligence regulatory functions and the development of industrial innovation. At the level of general education, it fosters artificial intelligence literacy as a Security Valve of regulation to mitigate risks associated with artificial intelligence, and realizes the fair, harmonious and innovative future development of artificial intelligence.
Key words : artificial intelligence regulation; European Union; Artificial Intelligence Act; artificial intelligence risk
引言
隨著人工智能系統普遍嵌入數字經濟產業，其引發的算法偏見、數據隱私、倫理失范等潛在風險也接踵而至。正如聯合國教科文組織（United Nations Educational, Scientific and Cultural Organization, UNESCO）在《人工智能倫理建議書》所說，人工智能可以對人類大有裨益并惠及全球，但也可能加劇偏見而導致算法歧視、數字鴻溝和數字互斥，威脅文化、社會和生物多樣性并造成社會經濟失衡。與此同時，該文件將人工智能監管視為人工智能風險的解決之道，鼓勵各國政府采用人工智能監管（Artificial Intelligence Regulation，AI Regulation）以預測后果、減少風險、避免有害后果、促進公民參與和應對社會挑戰［1］。在此背景下，歐盟逐步意識到人工智能風險引起的監管需求，構建起人工智能監管范式并創新政策體系和實踐舉措，逐漸成為世界人工智能監管領域的重要一極。
