从动物到AI——对意义理解诸形态的现象学考察
作者简介:高松，同济大学哲学系副教授（上海200092）。
摘要: 从现象学的视角看，动物、人类以及以GPT为代表的大语言模型在意义理解方面存在区别。动物通过形成类型来理解其环境，这是前语言的意义形态。语言的出现让人类拥有了概念和语境等高级意义形态，通过概念和语境的相互成就，创造了意义更为丰富的世界。GPT拥有建基于数学模型的独特意义理解机制。但由于无法直接连接到作为意义之源的外部世界，目前GPT只能与人类现有的说话方式对齐，尚不能凭借其独特的意义机制独自进行意义的创造与更新。
From Animal to AI—— A Phenomenological Investigation into Various Forms of Meaning-understanding
Abstract: From a phenomenological perspective,there are differences in the understanding of meaning among animals,humans,and LLMs represented by GPT.Animals understand their environment by forming types,which is a pre-linguistic form of meaning.The advent of language has endowed humans with advanced forms of meaning,such as concepts and contexts,which,through their interplay,have birthed a richer world of meaning.Meanwhile,GPT possesses a unique mechanism for understanding meaning based on mathematical models.However,since it cannot directly connect to the external world as the source of meaning,GPT currently can only align with existing human ways of speaking and cannot independently generate or update meaning leveraging its unique meaning mechanism.
点击查看大图
下载:
沪公网安备 31010102003103号