【摘要】 以ChatGPT、Sora为代表的生成式人工智能（AIGC），因训练语料、人工标注等环节自带政治、性别、种族等偏见，其道德缺失问题已成为重要议题。AIGC所引发的价值观偏差与伦理失范困境，更是关乎人类未来发展的关键挑战。文章运用技术现象学还原法，立足AIGC现实应用场景与道德属性，剖析其内在道德困境。在此基础上，对技术实践中的伦理争议展开批判性反思，构建技术伴随式伦理框架。进而探讨将道德原则嵌入智能系统设计的三条治理路径，即自上而下式进路、自下而上式进路以及自上而下与自下而上相结合的混合式进路，以期为破解AIGC道德困境提供新思路与实践方略，推动其走向合规、稳健且可持续的发展道路。更多还原

【Abstract】 Generative artificial intelligence(AIGC),exemplified by ChatGPT and Sora,has become a major topic of concern due to ethical shortcomings stemming from inherent biases—such as political,gender and racial biases—in its training corpora and manual annotation processes. The value distortions and ethical lapses triggered by AIGC represent critical challenges for humanity’s future development. Employing a phenomenological reduction method,this study analyses the intrinsic ethical dilemmas of AIGC by examining its practical applications and moral attributes. Building upon this,it engages in critical reflection on ethical controversies within technological practice to construct a technology-accompaniment ethical framework. We then explore three governance pathways for embedding ethical principles into the design of intelligent systems:a top-down approach,a bottom-up approach,and a hybrid approach combining both top-down and bottom-up elements. The aim is to provide new insights and practical strategies for resolving the ethical dilemmas of AIGC,thereby steering it towards a path of compliant,robust and sustainable development.更多还原