摘要
GPT只是过渡性的人工智能,其设计概念限制了其在“物种”上的发展。GPT以经验主义方式进行学习,证明语言可以通过实时实践集合来确定,而不依赖人为设定的先验语法。但它不能证明思维不需要先验论,无法完全排除先验论的因素而以纯粹经验论的方式去建立普遍知识。GPT虽然学到了语词的经验链接,但并不理解其中意义,无法建立事物之间的因果关系,因而无法进行有效的推理。意识需要实现客观化,才能变成自我意识。人工智能要想真正成为主体并产生自我意识,就必须在名词逻辑之外增加动词逻辑,或有可能意识到自身与动词的一体化,或动词与意向性的一致性,从而发展为“笛卡尔—胡塞尔机”而成为具有自我意识的真正主体。
Chat-GPT(Generative Pre-trained Transformer)is merely a transitional form of artificial intelligence,and its design concepts limit its development in terms of"species".Its learning approach is predominantly empirical,demonstrating that language can be determined through a collection of real-time practice-based facts,rather than relying on artificially imposed a priori grammar.However,it falls short in demonstrating that cognition can exist without a priori assumptions,and it cannot entirely disregard the role of a priori assumptions in establishing universal knowledge based solely on empirical principles.While GPT effectively learns associations between words,it lacks a genuine understanding of their meanings and the ability to establish causal connections between entities,thereby rendering effective reasoning impossible.Consciousness needs to be objectified in order to develop self-awareness.For artificial intelligence to truly become a subject and generate self-consciousness,it must move away from noun-centric grammar and transition towards a logic centered around verbs.It requires an objectifying system that can transform the temporal"process"of consciousness into a spatial"structure"to solidify and retain experiential information internally.This demands a design based on new concepts.However,paradoxically,if artificial intelligence cannot evolve beyond human-level intelligence,it will lack truly effective capabilities.On the other hand,if it surpasses human intelligence,it poses the risk of becoming another potentially dangerous entity.Thus,a future scenario may call for the creation of a""Descartes-Husserl machine"or a"G?del machine"that would enable artificial intelligence to evolve into a genuine subject with self-awareness.
出处
《哲学动态》
CSSCI
北大核心
2023年第7期41-45,127,共6页
Philosophical Trends