中文屋介绍

Introduction of Chinese Room

1. 中文屋是什么?

What is Chinese Room?

中文屋是袁毓林教授研究团队的学术主页,主要发布袁毓林教授及其研究团队在汉语语言学、计算语言学与相关领域的研究成果与学术进展。网站系统展示了团队成员自1980年代至今取得的各项科研成果,包括学术著作、期刊论文、会议发表、科研项目,以及在北京大学和澳门大学支持下开发的多项应用资源与技术评测任务。

Chinese Room is the academic homepage of Professor YUAN Yulin and his research team, introducing their research outputs and scholarly developments in Chinese linguistics, computational linguistics, and related fields. The website offers a systematic overview of the research accomplishments attained by the team members since 1980s, including academic monographs, journal articles, conference presentations, and research projects, as well as a range of application resources and shared tasks developed with the support of Peking University and the University of Macau.

2. “中文屋”名称的由来

Where does the concept of Chinese Room originate from?

“中文屋”这个名字,来自哲学史上一个著名的思想实验。

1980年,美国哲学家约翰·塞尔(John Searle)在Minds, brains, and programs中提出了一个设想:一个完全不懂中文的英国人被关在一间屋子里,屋外的人不断传入中文纸条,他手边有一本详细的操作手册,告诉他收到某种符号时该回传哪种符号。照着手册操作,他传回的答案让屋外的人以为他完全懂中文——但实际上,他对中文一窍不通,只是在机械地匹配符号而已。

根据塞尔的见解,电脑就是这样工作的:它们无法真正理解接收到的信息,但是它们可以运行一个程序,处理信息,然后给人一种它们有智能的印象。塞尔的本意是通过这个思维实验,来推翻强人工智能(机能主义)提出的过强主张:只要计算机拥有适当的程序,理论上就可以说计算机拥有它的认知状态以及可以像人一样地进行理解活动。

但是,我们同意这样的观点:中文屋子里的人、手册等都不懂中文,但作为一个整体的这间屋子(人和手册)是理解中文的。因此,自然语言处理依赖于语言知识资源和语言处理系统的有机结合。

我们将这个网站命名为“中文屋”,正是希望以此为出发点,探讨语言、理解与智能之间的关系,并持续思考那个至今悬而未决的问题:当我们与AI对话时,它究竟是在理解我们,还是只是一间更复杂的中文屋?

The name "Chinese Room" comes from a famous thought experiment in the history of philosophy.

In 1980, American philosopher John Searle proposed the following scenario in his article Minds, brains, and programs: an English speaker with no knowledge of Chinese is locked inside a room. People outside continuously pass in slips of paper written in Chinese, and the person inside has a detailed instruction manual telling him which symbols to send back in response to each symbol he receives. By following the manual, his replies convince those outside that he understands Chinese perfectly — yet in reality, he has no understanding of Chinese whatsoever. He is simply matching symbols mechanically.

According to Searle, this is how computers work: they cannot truly understand the information they receive, but they can run a program that processes that information and creates the impression of intelligence. Searle's intention was to use this thought experiment to refute the strong claims of Strong AI (functionalism) — namely, that a computer running the right program could, in principle, be said to possess genuine cognitive states and to understand in the same way humans do.

However, we subscribe to the view that while neither the person inside the room nor the manual understands Chinese, the room as a whole — the person and the manual together — does. It follows that natural language processing depends on the organic integration of linguistic knowledge resources and language processing systems.

We named this website "Chinese Room" as a starting point for exploring the relationship between language, understanding, and intelligence — and to keep asking a question that remains unanswered to this day: when we converse with AI, does it truly understand us, or is it simply a more sophisticated Chinese Room?

3. 中文屋 Logo 及其设计理念

Logo Design Concept

这个标识用英国数学家、计算机科学之父艾伦·图灵的名字组成图案:T、U 组成屋顶、窗口和一面墙壁,L 组成屋子的另一面墙壁,I 变形成为回头看书和伸手向窗外传递消息的小人,N、G 是叠放着的书籍和词典,也就是小人回头看的中文词典和语法书籍。

整个标识表示基于图灵测试原理的塞尔提出的“中文屋”思维实验。

The logo is composed of the letters in the name of Alan Turing — the British mathematician and father of computer science. The letters T and U form the roof, window, and one wall of the room; L forms the opposite wall; I is stylized into a small figure looking back at a book while reaching out through the window to pass a message; and N and G represent stacked books and dictionaries — the Chinese dictionary and grammar books the figure is consulting.

The logo as a whole represents Searle's "Chinese Room" thought experiment, based on the principles of the Turing Test.

中文屋 logo 设计图