摘要
从技术伦理的角度看,人工智能大模型存在开源与闭源两种发展模式,各自呈现出独特的优势与局限性。通过系统分析人工智能开源模型训练中的伦理风险,指出其在数据隐私、模型攻击、技术垄断及算法歧视等方面存在显著挑战。开源模式虽推动技术创新,但数据未经授权使用、恶意代码注入、算法偏差等问题可能加剧社会不公。提出协同治理框架,强调提升透明度、保障数据质量及伦理先行原则,倡导通过多方共治平衡技术开放与风险防控。
From a techno-ethical perspective,large AI models exhibit two distinct development models,open-source and closed-source,each demonstrating unique advantages and inherent limitations.Through systematic analysis of ethical risks in the training of open-source AI models,this study identifies significant challenges in areas such as data privacy,model attacks,technological monopolies,and algorithmic discrimination.Although open-source models promote technological innovation,issues such as unauthorized data use,malicious code injection,and algorithmic discrimination may exacerbate social injustice.This study proposes a collaborative governance framework,emphasizing the enhancement of transparency,data quality assurance,and the principle of ethical priority,advocating for multi-stakeholder co-governance to balance technological openness and risk prevention.
作者
闫明
龙丽达
Yan Ming;Long Lida(Jilin Normal University,Siping 136000)
出处
《西部学刊》
2025年第13期9-12,共4页
Journal of Western
基金
2025年度吉林省研究生教育教学改革课题“研究生‘大思政课’实践育人模式创新研究”(编号:2024199EX2L003O)的有关成果。
关键词
人工智能
开源模型
伦理风险
数据隐私
算法歧视
AI
open-source models
ethical risks
data privacy
algorithmic discrimination
