生成式人工智能的伦理风险及其防范对策  

Ethical Risks of Generative Artificial Intelligence and Preventive Strategies

作　　者：薛桂波 XUE Guibo(School of Marxism,Nanjing Forestry University)

机构地区：[1]南京林业大学马克思主义学院

出　　处：《南京林业大学学报(人文社会科学版)》2025年第1期1-11,共11页Journal of Nanjing Forestry University(Humanities and Social Sciences Edition)

基　　金：国家社会科学基金项目“基于中国语境的生成式人工智能伦理风险研究”(24BZX108);江苏省社会科学基金重点项目“生成式人工智能的伦理风险及治理对策研究”(23ZXA001);江苏高校哲学社会科学研究重大项目“江苏省科技创新的伦理治理体系构建研究”(2023SJZD114)。

摘　　要：作为人工智能技术发展的最新阶段,生成式人工智能虽然在创造性、交互性和通用性等方面展现出不同于传统人工智能的独特性,但是它的创新应用具有极高的不确定性,潜藏着巨大的伦理风险,诸如传播偏见和歧视、泄露隐私、制造虚假信息、操纵认知和导致科研伦理失范等。笔者认为应在分析生成式人工智能伦理风险的技术逻辑和社会建构等根源的基础上探寻生成式人工智能伦理风险的防范之策,强化“增进人类福祉”的价值理念、增强技术可解释性、健全伦理规制、加强企业自律和构建风险治理的全球合作机制等。As the latest stage in the development of artificial intelligence(AI)technology,generative AI exhibits distinct characteristics compared to traditional AI in terms of creativity,interactivity,and generality.However,its innovative applications come with high uncertainty and pose significant ethical risks,such as spreading bias and discrimination,infringing on privacy,generating false information,manipulating cognition,and leading to ethical misconduct in scientific research.The author believes that preventive strategies for the ethical risks of generative AI should be explored based on an analysis of their roots in both technological logic and social construction.These strategies should include strengthening the value concept of“enhancing human well-being”,improving the explainability of technology,establishing sound ethical regulations,reinforcing corporate self-discipline,and building a global cooperation mechanism for risk governance.

关 键 词：生成式人工智能 伦理风险 科技向善 科技伦理治理 

分 类 号：TP1[自动化与计算机技术—控制理论与控制工程]