人工智能的自主性与责任

Autonomy and Responsibility of Artificial Intelligence

目前,学界对人工智能是自主的且能够独立承担责任的论证都是基于行动自主的,这并不能解决人工智能承担责任的问题。人工智能的自主性问题实质上是在决定论世界里讨论能动者能否自主的问题。在理性回应论的支持下,人工智能有可能是完全自主的,并且能够承担义务责任。人工智能所承担的义务责任是一种联合义务责任,根据与人类关联的形式,分为多种责任承担形式。人工智能的使用者需要注意忽视的责任,而非人工智能的使用者要承担不使用人工智能的责任。

Currently,the argument that artificial intelligence(AI)is autonomous agent and bearer of responsibility is based on the autonomy of action,which cannot sustain AI to taking responsibility. The essence of whether AI is autonomous is about whether an agent can be autonomous in a deterministic world. AI can be autonomous,even a bearer of liability responsibility,if supported by reasons-responsive accounts.The liability responsibility attributed to AI is a kind of joint responsibility. It has different forms depending on how AI cooperates with humans. Users of AI should pay attention to the omission of responsibility. Meanwhile,nonusers of AI may take responsibility for not engaging AI as well.