首页|大模型在公共行政应用中的伦理风险与规制路径

大模型在公共行政应用中的伦理风险与规制路径

扫码查看
大模型作为人工智能发展前沿,具备理解与创新的卓越性能,正被广泛应用于公共行政场景当中,包括为行政人员提供数智决策参考、辅助探索智慧治理模式、优化政府内部管理.除了给公共行政带来更为颠覆性的变革,大模型可能会加剧传统的技术伦理问题并引发新的伦理风险,表征为安全侵犯、公正减损、信任破坏、责任不明、人的自主性消解等样态.这些伦理风险的形成原因涉及双重维度,既包括内源性的技术因素,如技术内置的涌现性、生成性与同质性,又包括外源性的人的因素,如行政人员的认知局限、不当商业竞争、规制滞后.立足于伦理风险表征与成因分析,克服技术缺陷、重塑个人认知、调整应用场景、强化监管举措成为规制伦理风险的现实选择,进而使大模型应用的有效性与持续性显著提升,并赋予公共行政数字化转型良性色彩.
Ethical Risks and Regulatory Paths in the Application of Big Models in Public Administration
As the frontier of AI development,big models,with excellent performance in understanding and innovation,are being widely used in public administration scenarios,including providing intelligent decision-making references for executives,assisting in exploring intelligent governance models,and optimizing internal government management.Besides bringing about more disruptive changes in public administration,big models may exacerbate traditional techno-ethical problems and cause new ethical risks,characterized by patterns of security violations,justice erosion,trust destruction,uncertainty of responsibility,and dissolution of human au-tonomy.The causes of these ethical risks can be found in a dual dimension,including both endogenous techno-logical factors and exogenous human factors.The former may include the emergence,generation and homogenei-ty of technologies,and the latter may include the cognitive limitations of executives,adverse business competi-tion,and regulatory lag.Based on the characterization of ethical risks and analysis of their causes,overcoming technical deficiencies,reshaping individual perceptions,adjusting application scenarios,and strengthening reg-ulatory initiatives have become realistic choices for regulating ethical risks,which in turn significantly improves the effectiveness and continuity of the application of big models and gives a benign trend to the digital transfor-mation of public administration.

big modelethical riskspublic administrationtechnological ethicsartificial intelligence(AI)

李旸

展开 >

湘潭大学 公共管理学院,湖南 湘潭 411105

大模型 伦理风险 公共行政 技术伦理 人工智能

2024

昆明理工大学学报(社会科学版)
昆明理工大学

昆明理工大学学报(社会科学版)

CHSSCD
影响因子:0.676
ISSN:1671-1254
年,卷(期):2024.24(6)