摘要
文章对从网络问答社区代表知乎和ChatGPT上同一提问获得的人工高赞答案和机器回答开展内容分析,基于外部特征、内部特征、情感与认知等探讨机器回答相对于人工问答的优劣。研究发现:ChatGPT能使人更便捷地获得信息,回答特征接近人工认可的高赞回答,但针对不同主题的回答质量存在差异,也存在虚假信息;以ChatGPT为代表的结合人工智能技术生成内容驱动的工具将广泛应用于人类活动,但需要防范信息误导、认知对抗、文明失控等风险。
This article conducts a content analysis of human and machine responses to the same question from the online Q&A communities Zhihu and ChatGPT,and explores the advantages and disadvantages of machine responses over human responses based on external features,internal features,emotions,and cognition.The study suggests that:ChatGPT allows people to have a easier access to information,with response characteristics close to human-approved,highly liked answers,but there are differences in the quality of responses provided by ChatGPT for different topics,as well as disinformation.In the future,ChatGPT-like tools driven by AI-generated content(AlGC),will be widely used in human activities,but risks such as misinformation,cognitive confrontation and loss of civility need to be guarded against.
作者
施亦龙
许鑫
SHI Yilong;XU Xin
出处
《图书馆论坛》
CSSCI
北大核心
2024年第1期151-159,共9页
Library Tribune