人民论坛·学术前沿2024,Issue(14) :31-44.DOI:10.16619/j.cnki.rmltxsqy.2024.14.004

建构文化人工智能的可能路径

Possible Paths to Constructing Cultural Artificial Intelligence

魏屹东
人民论坛·学术前沿2024,Issue(14) :31-44.DOI:10.16619/j.cnki.rmltxsqy.2024.14.004

建构文化人工智能的可能路径

Possible Paths to Constructing Cultural Artificial Intelligence

魏屹东1
扫码查看

作者信息

  • 1. 山西大学哲学学院
  • 折叠

摘要

生成式人工智能大模型的发展表明,新一代人工智能通达通用人工智能的目标正在接近.而通用人工智能要实现通用性、人性化和广泛应用,具有人类文化特征的属性是其必备的要件.将意识、情感、伦理等高级认知因素嵌入人工智能系统,也就是建构具有人性的文化人工智能,无疑是一种可能出路.如何让人工智能成为文化人工智能这种具身人工智能是一个重大挑战.从哲学方法论审视,建构文化人工智能的可能路径包括,智能体的范畴论与语境论的整合,感性与理性的整合,通过机制进行功能建模,通过结构功能实现适应性表征,以及通过语境觉知应对意外情况,最终通过严格的"图灵测试".如果这些都做到了,就有望在人工智能系统中嵌入人类特有的文化特征,使其成为具有人性的文化人工智能.

Abstract

The development of large language models of Generative Artificial Intelligence(GAI)suggests that a new generation of AI is approaching the goal of Artificial General Intelligence(AGI).In order to realize generality,humanization and wide application of general artificial intelligence,it is necessary to have the attributes of human cultural characteristics.Embedding high-level cognitive factors such as consciousness,emotion and ethics into artificial intelligence systems,that is,constructing cultural artificial intelligence with humanity,is undoubtedly a possible way out.How can AI become embodied AI such as cultural AI is a major challenge.From the perspective of philosophical methodology,the possible paths to construct cultural AI include the integration of category theory and contextual theory,the integration of sensibility and rationality,functional modeling through mechanisms,adaptive representation through structural functions,and response to unexpected situations through contextual awareness,and finally passing the rigorous"Turing test".If these are done,it is expected to embed human-specific cultural characteristics in the AI system,making it a cultural AI with humanity.

关键词

大语言模型/通用人工智能/具身人工智能/文化人工智能/适应性表征

Key words

big language model/artificial general intelligence/embodied artificial intelligence/cultural artificial intelligence/adaptive representation

引用本文复制引用

基金项目

国家社会科学基金重大项目(21&ZD061)

出版年

2024
人民论坛·学术前沿

人民论坛·学术前沿

CSSCI北大核心
ISSN:
参考文献量5
段落导航相关论文