生成式人工智能与知识生产

Generative AI and Knowledge Production

生成式人工智能与知识生产范式的变革是目前广泛关注的问题。文章首先对质疑人工智能不能进行知识生产的观点进行分析与回应,在此基础上回顾知识媒介历史,探讨生成式人工智能知识生产的特点,总结得出三种知识媒介,即追求全面图书馆藏的“巴别图书馆1.0”、运用非线性方式组织知识网络的“巴别图书馆2.0”、归纳知识底层逻辑重新生成新书的“巴别图书馆3.0”。目前的生成式人工智能属于“巴别图书馆3.0”,它打破了传统的知识搜索加推理的知识生产方式,通过模仿人类知识进行知识重构。这种新的知识生产方式虽基于统计学,是海德格尔意义上的“新常人”产物,但也具有创造性。人机协作的知识生产是生成式人工智能具有创造性的重要条件,但这种新的知识生产方式并非传统意义上的“思”,会引发人类思维逻辑机器化,以及知识权威和知识标准混乱等潜在后果。

Generative AI and revolution in the paradigm of knowledge production are inspiring widespread concern and discussion currently.This paper first analyzes and responds to the argument that AI is incapable of knowledge production,and then explores the characteristics of generative AI knowledge production through a review of the history of knowledge media.The paper argues that there are three different knowledge media,namely,"Library of Babel 1.0",which pursues a comprehensive library collection;"Library of Babel 2.0",which uses a non-linear approach to organize knowledge networks;and"Library of Babel 3.0",which re-generates new books by summarizing the underlying logic of knowledge.The current generative AI is in the phase of the"Library of Babel 3.0",which breaks the traditional knowledge production method of knowledge search plus reasoning and reconstructs knowledge by imitating human knowledge.This new way of knowledge production is based on statistics,a production of Heidegger's"new Das Man",but it can be creative in eyes of human beings.The knowledge production of human-machine collaboration is an important condition for the creativity of generative artificial intelligence,but this new way of knowledge production is not"thinking"in the traditional sense,and it will lead human to think like machine,as well as creating the potential outcome such as the confusion of knowledge authorities and knowledge standards.