适应性表征是人工智能发展的关键  

Adaptive Representation Is the Key to the Development of Artificial Intelligence

作　　者：魏屹东[1] Wei Yidong

机构地区：[1]山西大学哲学社会学学院

出　　处：《学术前沿》2019年第21期29-37,共9页Frontiers

基　　金：国家社会科学基金重点项目“科学认知的适应性表征研究”阶段性成果，项目编号：16AZX006

摘　　要：人工智能的表征方式应该是适应性的,这是通过生物特别是人类适应环境到人造的机器适应环境的一个类比逻辑地推出的。如果我们承认生物是进化适应性的,那么就应当承认其认知表征能力也是适应性的'进而作为人造物的人工智能包括机器人,其认知表征也自然是适应性的,即适应性的主体会创造或产生适应性的结果。这一适应性转换的过程是通过人为设置语境实现的,语境的设置是通过建构语料库进行的,人工智能的适应性表征正是基于人为设置的语境进行的。因此,适应性表征就成为智能机器人能否像人那样行动的一个重要判据。The representation of artificial intelligence(AI)should be adaptive,which is logically derived from an analogy between the adaptation of living things,especially human beings,to the environment,and the adaptation of artificial machine to the environment.If we admit that organisms are evolutionarily adaptive,then we should admit that their cognitive representation ability is also adaptive.Furthermore,as for the man-made AI,including robots,their cognitive representation is also naturally adaptive,that is,the adaptive subject will create or produce adaptive results.This process of adaptive transformation is realized by setting the context,and the context is set by constructing a corpus;the adaptive representation of AI is based on the artificially set context.Therefore,adaptive representation becomes an important criterion for intelligent robots to act like humans.

关 键 词：人工智能 语境设置 适应性表征 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]