哲学家会被取代吗?——从胡塞尔现象学角度看ChatGPT等大型语言模型的发展  

Will Philosophers Be Replaced:A Discussion of the Development of ChatGPT and Other Large Language Models from the Perspective of Husserlian Phenomenology

作　　者：吴程程 WU Chengcheng(Qingdao Research Institute of Humanities and Social Sciences,Shandong University,Qingdao,Shandong,266237)

机构地区：[1]山东大学人文社科青岛研究院,山东青岛266237

出　　处：《自然辩证法通讯》2025年第3期49-56,共8页Journal of Dialectics of Nature

摘　　要：以ChatGPT为代表大语言模型最近获得了重大进展,它由于在生成文本方面的出色表现而获得了人们的关注,也引起了哲学研究者们的担忧。ChatGPT的基本架构决定了它的工作原理是以统计为基础的,这使它遭到了很多批评,其中主要的批评是:只随机处理数据而没有涉及世界模型、缺乏创造性以及大数据主义的路线错误。但从胡塞尔现象学的角度来看,ChatGPT表现出的缺陷另有原因。ChatGPT面临的挑战是如何建立一个可生成、能流动的世界模型和时间结构,它所运用的大数据主义路线,从胡塞尔的视域理论角度来说也可以获得支持。Large language model,represented by ChatGPT,has recently gained significant momentum,not only gaining attention for its outstanding performance in generating text,but also raising concerns among philosophical researchers.The basic architecture of ChatGPT dictates that it works on a statistical basis,which has led to a number of criticisms,the main of which are that it only randomly processes data without involving a world model,that it lacks creativity,and that it has a wrong way of Big Dataism.However,from a Husserlian phenomenological point of view,ChatGPT exhibits flaws for other reasons.The challenges for ChatGPT are the need to solve the problem of building a generative and fluid world model and building the temporal structure.Moreover,way of Big Dataism it employs can also be supported from a Husserlian perspective of the theory of horizon.

关 键 词：ChatGPT 世界模型 生活世界 视域 大数据路线 

分 类 号：TP180225[自动化与计算机技术—控制理论与控制工程]