人工智能刑事责任主体资格的审视

Examination of the Qualifications of Criminal Liability Subjects of Artificial Intelligence

摘要: 人工智能的发展引发了学术界对其刑事责任主体资格的思考。综合人工智能在发展的不同阶段中呈现出的特点，可以将人工智能从整体上分为弱人工智能和强人工智能。弱人工智能因其程序和逻辑的形成依赖于人类活动而不具有独立意识和自主决策能力，在更多情况下表现为辅助性质的工具。而强人工智能则具备更强大的学习能力，其能够在人类所设计程序以外的范围凭借自己的意志做出决定并以此行动，具有一定程度的自主行为能力。基于人工智能技术水平持续进步的趋势，不断有学者提出了人工智能可能在未来会产生和人类一样，甚至是超越人类的思维能力和自主意识。学界对人工智能刑事责任主体资格的认定主要表现为肯定论和否定论。综合人工智能发展程度的不确定性、法律内在的逻辑性等多方面因素，笔者认为肯定论的观点实际上缺少足够科学的论证而不易形成有效结论。

Abstract: The development of artificial intelligence has triggered academic thinking about its qualifications as a criminally responsible subject. The characteristics of integrated artificial intelligence in different stages of development can be divided into weak artificial intelligence and strong artificial intelligence as a whole. Weak artificial intelligence, because its programs and logic depend on human activities without independent consciousness and autonomous decision-making ability, is more often manifested as a tool of an auxiliary nature. Strong artificial intelligence, on the other hand, has a stronger learning ability, and can make decisions and act accordingly on its own will outside the scope of human-designed programs, with a certain degree of autonomous behavior. Based on the trend of continuous progress in the level of artificial intelligence technology, some scholars have proposed that artificial intelligence may produce the same thinking ability and autonomous consciousness as humans in the future. The academic community’s determination of the qualification of the subject of criminal responsibility of artificial intelligence is mainly manifested as the theory of affirmation and the theory of negation. Considering the uncertainty of the degree of development of artificial intelligence, the inherent logic of the law and other factors, the author believes that the affirmative view actually lacks sufficient scientific arguments and is not easy to form effective conclusions.