人工智能应用的界限及其规制:以保护人的尊严为底线  

The Boundaries and Regulation of Applications of Artificial Intelligence Based on Human Dignity

作　　者：黄文婷 Huang Wenting

机构地区：[1]武汉大学法学院

出　　处：《比较法研究》2025年第2期32-46,共15页Journal of Comparative Law

摘　　要：人工智能应用带来了人类能在多大程度上容忍其改写自身主体性地位和价值的现实问题。人工智能应用应当以人的尊严为底线,然而,该底线面临着三重风险,即人类在物种层面的支配性、公民在社会层面的主体性以及个人在个体层面的自主性受到挑战。为保护人的尊严,应当绝对禁止将个人当成纯粹的手段伤害且不会对个人、社会和人类产生任何福祉的人工智能应用,相对禁止对人的尊严产生风险但也会带来福祉的人工智能应用。在相对禁止的路径下,应当以人工智能应用侵犯人的尊严的风险程度划分不同的人工智能类型,同时采用比例原则进行比较权衡,以划定人工智能应用的界限。具体而言,法律应当严格规制带来重大风险的特定人工智能应用,合理规制带来有限风险的其他人工智能,同时法律有必要将保护人的尊严底线转化为应用者的主体性义务,并强调静态的合比例论证与动态的风险控制。The proliferation of applications of artificial intelligence(Al)raises fundamental questions about the extent to which human beings will tolerate Al-induced alterations to their subjective agency.While AI development should never compromise human dignity,contemporary implementations threaten three critical dimensions of this fundamental value:species-level supremacy,societal citizenship,and individual autonomy.This paper proposes a dual-prohibition regulatory paradigm to address these challenges.Absolute prohibition should apply to AI systems that instrumently exploit individuals without generating compensatory benefits for personal,social,or human welfare.Conditional restrictions should govern technologies presenting dignity risks alongside potential benefits.Under this risk-stratified framework,AI applications must undergo categorical classification based on their dignity infringement potentials.A proportionality assessment should then be applied to delineate operational boundaries through rigorous benefit-risk balancing.Specifically,legal mechanisms should:(i)impose stringent regulations on high-risk Al systems with substantial dignity implications;(i)implement measured governance for technologies that pose contained risks.Additionally,this paper advocates for legal reforms to codify fundamental human dignity protection into applicator's binding obligations.Regulatory compliance should be ensured through synergistic implementation of static proportionality evaluations and dynamic risk control mechanisms.

关 键 词：人的尊严 人工智能应用 比例原则 法律规制 保护义务 

分 类 号：D90[政治法律—法学理论]