ChatGPT智能处理器使用我们的语言吗?  

Does ChatGPT Use Our Language?

作　　者：周柏乔 CHOW Pak Kiu

机构地区：[1]香港公开大学,中国香港

出　　处：《哲学分析》2024年第4期3-16,196,共15页Philosophical Analysis

摘　　要：有人说ChatGPT智能处理器只会吐出话语,不(会)说话。尽管它的话有意思,而且可以被理解,但理解它的话真的在于挖掘它的“心思”或“心意”吗?假定与其对话的自然人都认为ChatGPT没有“心思”或“心意”,却发现对话有意思,起码有助于一般人通过ChatGPT的话语进一步认识自己的“心思”或“心意”,这是否意味着ChatGPT是个语言使用者,和人类的语言使用者一样,可通过与人们的对话帮助自然人认识自己的想法?如果是这样,就会出现矛盾:它不(会)说话却成为语言使用者。借用蒯因的翻译理论、戴维森的诠释理论以及维特根斯坦的后期论述,可以说明自然人有哪些语言本领,AI工程当予以实现,ChatGPT智能处理器方可媲美自然人的语言使用能力。If the outputs of ChatGPT do not reveal beliefs and desires,as what the human language is supposed to do,what could be left for these outputs to say?In this paper,I shall argue that ChatGPT can be taken as a language user that can do translation as quins expects,or do interpretation as Davidson anticipates,on condition that it can engage a dialogue with a human who is able to take its words meaningfully via his translation or interpretation.Nevertheless,given that ChatGPT does not have the human form of life,whatever it says about itself should not be taken seriously.There are some other shortcomings for ChatGPT too.It may go wide in doing radical interpretation when it is asked to rely as little as possible on established knowledge for some interpretations on a new subject,as ChatGPT has little use when it does not have to make use of the vast source that it has reserved.

关 键 词：ChatGPT 蒯因 戴维森 维特根斯坦 

分 类 号：B15[哲学宗教—世界哲学]