摘要
数字社会已经成为触手可及的现实,算法辅助决策的问题已经亟待解决。算法辅助决策较人类决策具有特殊的技术优势,所以得到很多人的自主、自愿选择。然而,算法参与人类意思的程度并非一致,可以由低到高区分为检索过滤算法、排序精选算法、偏好模拟算法和家长式算法。算法辅助决策亦存在风险,其复杂性使对其解释与认知越来越难,其设计者、运营者具有独立利益需求,其会偏离人的偏好,人的决策能力会受到算法辅助的影响。但是,总体上,算法辅助决策更具效率,更有助于增进人的幸福感,虽然在某些方面有损于人的尊严。据此考量,应当在承认算法超越人的意思能力的基础上,描绘评价算法意思参与的“智能性”光谱,构建“算法规制算法”的治理与保护路径,划分算法辅助决策的责任机制。
Digital society has become a tangible reality,and the resolution of the problem of algorithm-assisted decision making is already urgent.Algorithm-assisted decision making has been applied voluntarily by many people because it has special technical advantages over human decision making.However,the degree of involvement of algorithms in human will is not uniform and can be distinguished from low to high as retrieval-filtering algorithms,sorting-selecting algorithms,preference simulation algorithms and paternalistic algorithms.There are also risks in algorithmassisted decision making:its complexity makes its interpretation and perception increasingly difficult;its designers and operators have independent interest needs;its preferences can deviate from human preferences,and human decision making ability can be affected by algorithmic assistance.However,in general,algorithm-assisted decision making is more efficient and contributes more to human feeling of well-being,although in some ways it is detrimental to human dignity.Based on this consideration,the spectrum of"intelligence"for evaluating the meaningful participation of algorithms shall be drawn,the path of governance of"algorithms by algorithms"shall be constructed,and the mechanism of responsibility for algorithm-assisted decision making shall be made on the basis of recognizing that the will ability of algorithms exceed that of human.
出处
《东方法学》
CSSCI
北大核心
2022年第3期33-42,共10页
Oriental Law
关键词
算法
算法辅助决策
算法自动化决策
意思自治
数据
算力
algorithm
algorithm-assisted decision making
algorithm-automated decision making
autonomy of will
data
algorithmic power
