In the 1st issue of 2025 in the “Journal of Dialectics of Nature,” Xia Yonghong argues that the current dominant paradigm in the ethics governance of AI is value alignment, which aims to ensure that machine values are consistent with human values. This paradigm primarily employs representationalist and behaviorist AI solutions. However, they face significant challenges in accurately capturing and encoding complex human values due to the common sense problem. To address this issue, Xia proposes introducing embodied-generative AI solutions, which enable machines to better grasp the relevance in the world.
Xia further introduces the concept of “value symbiosis” as an alternative paradigm. This new paradigm aims to achieve a harmonious coexistence between machine values and human values. It encompasses two key AI design principles: mutual benefit in survival interests and mutual recognition of values. By adopting these principles, the value symbiosis paradigm seeks to foster a more integrated and dynamic relationship between human and machine values.
— Social Sciences Weekly, January 16