基于意向性视域下的人工智能符号主义与联结主义探究
Exploration of artificial intelligence symbolism and connectionism from the perspective of intentionality
谢佛荣 1韩小芬1
作者信息
- 1. 南华大学马克思主义学院,衡阳 421001
- 折叠
摘要
对于人工智能如何实现意向性一直是学界关注和研究的重要内容之一.针对如何跨越这一鸿沟,在人工智能的研究和实践发展中逐渐发展出了两种路径:一种是符号主义的模仿路径;一种是联结主义的还原路径.符号主义的模仿路径因符号系统的高度抽象化、符号化,使得这一路径并不能使人工智能完整具备人的意向性.联结主义虽然比符号主义的模仿路径具有一定优势,但联结主义的还原路径因人工智能在实现可解释性方面存在不足,所以这一方式并不能为人工智能具备人的意向性提供实质性帮助.要使人工智能像人类一样具备意向性,须更多整体考虑智能机器与外部世界环境的交互,把思维、计算与环境予以整体的融合.
Abstract
How to realize intentionality for artificial intelligence has always been one of the major concerns and researches in the academic world.In order to cross this gap,two paths have been gradually developed in the research and practical development of artificial intelligence:one is the symbolist imitation path;the other is the connectionist reduction path.Symbolism's imitation path is highly abstracted and symbolized by the symbol system,which makes this path not able to make AI complete with human intentionality.Although connectionism has some advantages over symbolism's imitation path,connectionism's reduction path can't provide substantial help for AI to have human intentionality due to the shortcomings of AI in realizing interpretability.In order for AI to be as intentional as human beings,it is necessary to consider more holistically the interaction between intelligent machines and the environment of the external world,and to integrate thinking,computing and the environment as a whole.
关键词
中文屋论证/意向性/人工智能/符号主义/联结主义Key words
Chinese house argument/intentionality/artificial intelligence/symbolism/connectionism引用本文复制引用
基金项目
国家社会科学基金后期资助项目(22FZXB095)
出版年
2024
NETL
NSTL
万方数据