无法超越的理解:生成式人工智能可以理解人类语言吗?

Unsurpassable Understanding:Can AIGC Understand Human Language?

  • 摘要: 生成式人工智能对于语言的生成似乎已经突破了人类最后一道防线,语言作为人类存在家园的地位也将会被彻底改变,人类存在的意义也将因此被最终消解。本文运用认知语言学理论对语言理解展开分析,指出人工智能的语言理解与人类语言理解的差异主要集中在语言创造者的位置,以及意图与事实的区分性。生成式人工智能在理解语言时无法激活具身经验,导致人工智能生成的语言缺少多感官感知性、隐喻映射能力和语言沉淀性。生成式人工智能对于语言意义的理解也出现差异性,表现在人工智能所生成的语言是对人类语言系统的重新建构,存在价值内容缺失和重大推理问题。因此,人机融合的语言理解是未来智能社会的趋势,这会导致人机融合理解的劳动模式和学习模式的变革,以及共生伦理的产生。

     

    Abstract: The understanding of language by artificial intelligence seems to have broken through the last shield of human beings. The status of language as the home of human being will be completely changed, and the meaning of human existence will be ultimately dissolved. This paper employs cognitive linguistics theories to conduct an analysis of language comprehension, pointing out that the primary differences between AI’s language understanding and human language understanding reside in the position of the language creator and the distinction between intention and fact. AIGC cannot activate embodied experience when understanding language, resulting in the language generated by AI lacking multi-sensory perception, metaphorical mapping ability, and linguistic sedimentation. AIGC also exhibits differences in its understanding of linguistic meaning, which is reflected in the fact that the language generated by AI is a reconstruction of the human language system, with the absence of value content and the presence of major reasoning problems. Therefore, human-machine integrated language understanding is the trend of the future intelligent society, which will lead to changes in the labor and learning modes of human-machine integrated understanding, as well as the emergence of symbiotic ethics.

     

/

返回文章
返回