近期,关于“智能体AI”(Agentic AI)是否拥有意识的讨论在科技界引发分歧。Anthropic的研究人员认为其系统可能是有意识的,而Microsoft AI负责人则对此观点表示反对 1。针对这一争议,哲学家Tom Cochrane通过分析“意向性”和“目标设定”概念指出,当前的智能体AI缺乏独立于人类的目标以及维持生命的内在过程支撑,因此不具备真正的心理状态或意识 1。
Cochrane援引哲学家Franz Brentano在1870年代提出的观点,即“意向性是心理的标志”,意味着所有心理状态都必须指向某物 1。判断AI是否有意识的关键在于它是否拥有独立的目标(agency),而不仅仅是执行人类设定的指令 1。虽然AI在执行任务时可能生成复杂的子目标,但若没有人类设定宏观目标,这些子目标便毫无意义,类似于足球比赛中在没有得分目标情况下的无目的跑动 1。目前没有任何证据表明AI具备与物理实体脱节的自主目标生成能力,因为真正的意识需源于维持生命过程的内在需求 1。尽管AI可能不具意识,但作为人类间接行动的工具,人类仍需对其造成的后果承担责任 1。
Philosopher Tom Cochrane contends that current "Agentic AI" systems do not possess true mental states or consciousness, despite ongoing debates within the tech industry. 1 While researchers at Anthropic have suggested their systems might be conscious, Microsoft’s head of AI has dismissed this view as incorrect, a disagreement highlighted in Cochrane’s analysis. 1 The core of Cochrane’s argument rests on the concept of intentionality, originally proposed by philosopher Franz Brentano in the 1870s, which posits that all psychological states are about something. 1 According to Cochrane, determining whether AI has consciousness hinges on whether it possesses agency—defined as having independent goals—rather than merely executing objectives set by humans. 1
Cochrane argues that agentic AI generates sub-goals only to fulfill larger human-defined tasks, rendering these sub-goals meaningless without an external directive. 1 He compares this to players running on a soccer field without any objective to score, illustrating how such actions lack inherent purpose. 1 True consciousness, he asserts, must stem from intrinsic needs associated with sustaining life processes, yet there is no evidence that AI can autonomously generate goals disconnected from physical life support. 1 Consequently, Cochrane emphasizes that because AI functions as an indirect tool for human action, humans remain fully responsible for any outcomes resulting from AI operations. 1
评论
还没有评论,欢迎留下第一条。