Abstract: With its wide range of application scenarios, ChatGPT can effectively empower product research and development, intelligent manufacturing, education, content production, etc., which has a far-reaching impact on promotion of industrial upgrading, development of social productivity, and improvement of production tools. However, ethical risks associated with rapid development of AI in the era of digital intelligence have aroused concern from all sides. As an important guarantee to ensure sustainable and healthy development of the science and technology activities of AI enterprises, the ethical governance of government in science and technology is of key importance to promote the resolution of social risks such as algorithmic bias and discrimination, data security challenges, and misuse of computing power, as well as to promote human well-being via science and technology innovation. Based on institutional theory, the study took listed AI enterprises on the main board of A-shares in China as the research object to test the effect and mechanism of the ethical governance of government in science and technology on science and technology for social good of AI enterprises. As the results have shown, (1) The ethical governance of government in science and technology as a formal institution is positively associated with science and technology for social good of AI enterprises; (2) Social trust and political connection as informal institutions play moderating roles in the relationship between the ethical governance of government in science and technology and science and technology for social good of AI enterprises, showing the overlapping effect of informal institutions in collaboration with the ethical governance of government in science and technology on driving science and technology for social good of AI enterprises; (3) The results of heterogeneity test indicate that the ethical governance of government in science and technology has a more significant effect on improving science and technology for social good of AI enterprises in samples with high regional Confucianism, large sizes, and low executive equity incentives. Based on above, the study provides new perspectives on driving corporate science and technology for social good in the era of digital intelligence and empirical insights for the government and enterprises to address ethical risks of AI and promote human well-being, bridging the gap of prior research that ignores the critical influence of science and technology ethics policies on corporate science and technology for social good in government-dominated formal systems. Moreover, from practical perspectives, the government should accelerate the process of the ethical governance of science and technology in the field of AI, strengthen the safety bottom line, promote AI enterprises to build explainable AI, and protect user privacy and data security. Meanwhile, the government needs to strengthen the construction of agile governance capacity, led by the government to create a dynamic linkage ecosystem with the participation of enterprises, academics, news media, and other diversified subjects, to enhance the government’s forward-looking governance level of ethical risks of AI. In response to the ethical governance of government in science and technology, which includes the principles of promoting human well-being and controlling risks reasonably, AI enterprises, under the guidance of regulations and regulatory policies, take the initiative to implement self-governance and self-regulation actions such as integrating ethics into corporate culture and values, incorporating ethical values into the design of products/services, and introducing ethical risk assessment mechanisms for AI products/services, so as to realize ethical business decision-making and take into account both short-term and long-term interests, thus promoting science and technology for social good.

摘要： 数智时代下人工智能高速发展所引致的伦理风险引发各方关注，政府科技伦理治理作为确保人工智能企业科技活动可持续发展的重要保障，对促进应对算法偏见与歧视、数据安全挑战与算力误用等社会性风险与依托科技创新增进民众福祉等科技向善行动具有关键意义。研究基于制度理论，将中国A股人工智能上市企业作为研究对象，考察政府科技伦理治理对人工智能企业科技向善的影响机制。结果显示：作为正式制度的政府科技伦理治理对人工智能企业科技向善具有显著的正向驱动作用；作为非正式制度的社会信任与政治关联在政府科技伦理治理与人工智能企业科技向善间起正向调节效应，呈现出非正式制度协同政府科技伦理治理对驱动人工智能企业科技向善具有叠加增益效果；异质性检验结果表明，政府科技伦理治理在区域儒家文化氛围高、企业规模大以及高管股权激励低的样本中，对人工智能企业科技向善的改善作用更明显。研究为数智时代驱动企业科技向善提供了新视角，也为政府与企业应对人工智能伦理风险、增进民众福祉提供经验启示。