电车难题与自动驾驶系统算法设计中的伦理考量

Trolley Problem and Ethical Considerations in Algorithm Design for Automated Vehicles

自动驾驶汽车是人工智能技术得到全面、综合应用的产业领域,涉及到算法自动化决策与人的自主性和尊严之间的潜在冲突。由于法律本身的回应性和救济性特征,导致它无法提前介入到算法设计过程之中,因而无法为算法设计提供向善的指引。我国虽然越来越重视人工智能伦理问题,并为此设立了相关机构,制定了相关规则,但关于人工智能伦理的讨论却停留在科技向善、以人为本等抽象原则的层面,没有形成有针对性的自动驾驶汽车算法伦理。本文通过对德国交通与数字基础设施部伦理委员会报告、伦理学中的电车难题讨论和道德心理学领域的“道德机器试验”的介绍和分析,讨论了为自动驾驶汽车装上伦理方向盘的必要性和可能的推进方式。

Automated vehicles incorporate sensors,positioning devices,decision-making algorithms and executors.They represent the most comprehensive application of artificial intelligence and raise sophisticated ethical and legal issues involving personal autonomy,human dignity and privacy.Due to the corrective and remedial nature of law,it cannot intervene in the design process in advance,therefore cannot provide a normative compass for algorithm design.Although China pays more and more attention to the ethical issues of artificial intelligence,and has established relevant institutions and formulated relevant rules for this purpose,the discussion on artificial intelligence ethics stays at the stage of proposing general principles such as technology for good and human-centered design,and yet to formulate a sector-specific set of ethical principles for automated vehicles.This article uses ethical principles proposed by the Ethical Commission of German Ministry of Transportation and Digital Infrastructure as the reference point,analyzing its strengths and weakness by related it to the trolley problem discussion in moral philosophy and the“moral machine”experiment in moral psychology.By doing this,it intends to suggest certain preliminary steps for China to develop its own AV-specific ethical principles and participate in the international dialogue on this increasing urgent and important issue.