【摘要】 “伦理即服务”以守则制定、伦理转译、效果审核的“三步走”机制实现了人工智能伦理原则的实践转化，成为一种具有平衡性、开放性和反思性的人工智能伦理治理新范式。以话语伦理理论和分布式道德责任理论为基础的“伦理即服务”理念创新了伦理转译方式，有助于保护多元化的价值观，为软道德提供作用环境，推动负责任人工智能的发展和普及。但也应看到EaaS存在着转译工具可用性不强、实际作用效果不确定、易引发新的伦理风险等局限。基于此，需监督管理EaaS的工具选择，有效评估EaaS的作用效果，积极应对EaaS的伦理风险，以克服EaaS存在的局限，切实推进人工智能伦理治理。更多还原

【Abstract】 Ethics as a service(EaaS) realises the practical transformation of AI ethical principles through a three-step mechanism of code formulation, ethical translation, and effect review, and thus qualifies as a new balanced, open, and reflective paradigm of AI ethical governance. Based on the theory of discourse ethics and distributed moral responsibility, EaaS innovates the way of ethical translation, helps to protect diversified values, provides an environment for soft ethics, and promotes the development of responsible AI. However, it should also be seen that EaaS has limitations such as weak usability of translation tools, uncertainty of practical effects, and triggering new ethical risks. Thus, it is necessary to supervise and manage the selection of tools for EaaS, effectively assess the effect of EaaS, and actively deal with the ethical risks of EaaS, so as to effectively advance AI ethical governance.更多还原