摘要
ChatGPT与人类用户的交流依赖于其内部语言模型,从具体的语义处理机制来看,人机交流表现出一种类“预测加工”的过程,在这种观点下,人机交流中的相互认知是一种双向预测的耦合。但人人交流中的社会认知并非仅仅依赖于内部机制运行,还依赖于人与人之间的交互作用,而交互作用具有独立性,不能还原到个体机制的解释框架之中,通过交互产生的认知结果具有具身性、关系性和涌现性。人人交流是一种参与式的意义构建过程,但是ChatGPT与人的交流并非具身,而是完全依赖内部模型机制的运行,交流过程仅仅是对人类赋予意义的单向识别,提供给用户的只是一种泛情景化、知识化和经验化的回复。人机交流想要走向社会认知,并非仅改进机器以单方面贴近人人交流的模式,而是人与机器之间的相互弥合,同时机器广泛参与到与人类的交流过程中必然会带来社会认知的内涵与边界的改变,因此应当在人类与机器的动态变化关系中为人机交流寻找出路。
The smooth communication between ChatGPT and human users relies on its internal language model,and in terms of specific semantic processing mechanism,the communication between ChatGPT and human beings shows a kind of“predictive processing”process,in which the mutual cognition in human-computer communication is a two-way predictive coupling.However,social cognition in human-human communication does not only rely on the operation of internal mechanism,but also on the interaction among humans.Moreover,the interaction is independent and cannot be explained in the framework of individual mechanism.Cognitive results produced by the interaction are embodied,relational,and emergent,which makes the human-human communication a participatory sense-making process.However,ChatGPT’s communication with human beings is not embodied,but completely relies on the operation of internal model mechanism,therefore,its communication process is only a one-way recognition of the meaning given by human beings,and what is provided to the users is only a kind of pan-situational,intellectualized and empirical response.If human-machine communication wants to achieve the effect of human communication,it is not enough to improve the machine unilaterally close to the mode of communication among humans,but also necessary to bridge the gap between humans and machines.The extensive participation of machine in the process of communicating with human beings will inevitably bring about a change in the connotation and boundaries of social cognition,so we should look for a way out for the human-machine communication in the dynamic change of the relationship between humans and machines.
作者
孙瑞璇
谭笑
SUN Ruixuan;TAN Xiao(College of Political Science and Law,Capital Normal University,Beijing 100089,China)
出处
《山东科技大学学报(社会科学版)》
2024年第1期17-25,共9页
Journal of Shandong University of Science and Technology(Social Sciences)
关键词
ChatGPT
人机交流
预测加工
社会认知
参与式意义建构
ChatGPT
human-computer communication
predictive processing
social cognition
participatory sense-making