人工智能符号接地问题研究的意义和挑战  

The Research Significance and Challenge of the AI Symbol Grounding Problem

作　　者：霍书全[1] HUO Shuquan

机构地区：[1]安徽大学哲学系

出　　处：《上海师范大学学报（哲学社会科学版）》2019年第3期98-107,共10页Journal of Shanghai Normal University(Philosophy & Social Sciences Edition)

基　　金：国家社科基金重大项目"面向自然语言理解的逻辑构建和符号接地问题的哲学;心理学研究"(18ZDA032)的阶段性成果

摘　　要：符号接地问题要解决符号如何获得意义、符号和它们的意义如何联系起来的问题。该问题由塞尔中文屋论证所引发,加拿大认知科学家哈纳德首先明确地给出表述并试图解决,历史上皮尔士曾经论及过类似的问题。要真正刻画人的智能就需要使符号接地,因此它有重要的研究意义。思蒂尔斯2007年给出了一个解决方案,之后塔迪欧和弗洛里迪给出了另一个解决方案,由此引起了一系列的批评和争论。迄今为止,所有的解决方法都不令人满意,因为这一问题会涉及很多跨学科的难题,要解决它似乎有很长的路要走。文章尝试为一些疑难问题指出可能的解决途径。尽管面临很多难题,但在解决符号接地问题的过程中,很多哲学和科学问题都会得到澄清或解决。The problem of symbol grounding is to solve the problem of how symbols acquire meanings and how symbols relate to their meanings. This problem was triggered by Searle’s Chinese House argument. Stevan Harnad, a Canadian cognitive scientist, first made a clear statement and tried to solve it. Charles Saunders Peirce also discussed the similar problem in history. In order to depict human intelligence, it is necessary to make symbols grounded, so it is of important research significance. Luc Steels offered one solution in 2007, and then Tardio and Floridi came up with another solution, thus prompting a flurry of criticism and debate. So far, none of the solutions have been satisfactory, as the problem involves many interdisciplinary difficulties, and there seems to be a long way to go to solve it. This paper attempts to point out possible solutions to some difficult problems. Despite many difficulties, in the process of solving this problem, many philosophical and scientific problems will be clarified or solved.

关 键 词：符号接地问题 人工智能 符号主义 认知科学 语言哲学 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]