摘要
用户与大模型应用交互的本质是人与"众人"互动,超越了"第二自我"所定义的交互场景.本文主要讨论以下四个方面的问题.第一,分析了从大众媒体公信力到大模型应用信任的四个发展阶段,从时间维度考察了大模型信任的含义.第二,讨论了基于大模型应用的信任内涵、外延与生成模式的变化与特点.第三,分析大模型应用信任带来的三种人机关系.第四,提出了基于大模型应用信任的三个值得思考的问题:人机伴侣关系中对大模型"众我的平庸"信任对个体认知的影响;用户对大模型的信任是否会加重用户网络化的个人中心主义;公众对大模型的信任是否会影响整体的社会资本.
Abstract
The essence of the interaction between human users and the LLM application is the interaction between the person and the"crowd",which is beyond the interaction scene defined by"the Second Self'.This paper first analyzes the four stages of development from mass media credibility to trust of LLM application from a historical dimension,proposing the meaning of trust of LLM.Second,we discuss the changes and characteristics of the connotation,extension,and generation mode of trust based on the LLM application.Third,we discuss the three kinds of human-machine relationships brought about by LLM application trust:tool-using,agent-based,and companion-based relationships.Finally,we raise three questions for consideration:whether the"mediocrity for the crowd"of LLM will impact individual perceptions in partner relationships,whether users'trust of LLM will exacerbate users'networked individualism,and whether the public's trust in LLM will affect the overall social capital.
NETL
NSTL
万方数据