预知性技术伦理消解AI科林格里奇困境的路径分析

Analysis on the Path of Resolving the Collingridge’s Dilemma of AI by Anticipatory Technology Ethics

人工智能的关键特征是采取最优行动,而该特征会直接导致人工智能难预知性和风险性的属性,进而使其社会负面影响呈现出早期难预测和后期难控制的特点,此称为人工智能的科林格里奇困境。而单一的技术向度解决科林格里奇困境对人工智能技术的特性提出了严苛要求,导致其适用范围的局限性。鉴于此,我们引入了现代伦理方法:预知性技术伦理(ATE),并用行动者网络理论对其责任分配问题进行了补充。在技术设计阶段渗透价值并映照社会行动者网络以解决人工智能社会后果在前期难预测、后果难纠正的问题,进而达到消解人工智能科林格里奇困境的目标。

The key feature of artificial intelligence is to take optimal actions,this feature will directly lead to the unpredictable and risky attributes and it will cause negative social impact which is difficult to predict in the early stage and difficult to control in the later stage.This is called the Collingridge’s Dilemma of Artificial Intelligence.A single technical solution to the Collingridge’s Dilemma places strict requirements on the characteristics of artificial intelligence technology,resulting in the limitation of its scope of application.In view of this,we introduce a modern ethical method:Anticipatory Technology Ethics (ATE),and use the Actor-network theory to supplement its responsibility assignment.Infiltrating value in the technology design phase,and referring to the social Actor-network to solve the problems artificial intelligence’s social consequences are difficult to predict in the early stage and difficult to correct in the later stage,then the goal of dissolving the Collingridge’s Dilemma of Artificial Intelligence is reached.