生成式人工智能的风险特征及语义图像:基于网络社区讨论的计算文本分析  

Risk Characteristics and Semantic Imagery of Generative AI:A Computational Text Analysis Based on Online Community Discussions

作　　者：匡恺 刘力铭 KUANG Kai;LIU Liming(School of Journalism and Communication,Tsinghua University,Beijing 100084,China)

机构地区：[1]清华大学新闻与传播学院,北京100084

出　　处：《河北大学学报（哲学社会科学版）》2024年第6期147-160,共14页Journal of Hebei University(Philosophy and Social Science)

基　　金：清新计算传播学与智能媒体实验室研究支持计划(2024TSLCLAB001)。

摘　　要：生成式人工智能在带来技术突破和效率提升的同时,也在公众视野内演变成了不可见的技术风险。研究立足于技术的风险语义图像视角,采用计算文本分析的方法,结合文本主题模型、词向量模型与语义共现网络,对网络社区(知乎)中的人工智能风险讨论进行分析。研究结果显示,公众对人工智能的风险感知围绕Chat GPT在信息传播中的短期风险和人工智能在人类社会中的长期风险。从风险特征维度看,公众风险感知呈现出个体化趋势,人工智能被视为一种自愿卷入的低危害风险,但在认知上引发了公众的不确定感,个体对风险类型的识别集中在国家安全风险和个体经济风险,呈现出宏观政治与微观个体利益相结合的认知框架。公众主要采取两类应对策略,分别是以知识为中心的自我学习策略和以行业发展为中心的共同应对策略,最终形成了技术调节、技术规制、生存适应、系统共振四种风险语义图像。While generative AI brings technological breakthroughs and improves efficiency,it has also evolved into an invisible technological risk in the public view.This research is grounded in the perspec-tive of the semantic imagery of technological risks and employs computational text analysis,integrating LDA topic modeling,word vector models,and semantic networks to analyze discussions on AI risks within online communities(Zhihu).Results show that public perceptions of AI risks focus on the short-term risks of ChatGPT in information dissemination and the long-term risks of AI in human soci-ety.Public risk perception exhibits an individualization trend,with AI being viewed as a low-harm risk of vol-untary involvement,while cognitively triggering public uncertainty.The identification of risk types by individ-uals is focused on national security risks and personal economic risks,presenting a cognitive framework that combines macro-politics with micro-individual interests.The public mainly adopts two coping strategies:a knowledge-centered self-learning strategy and an industry-centered collective response strategy,ultimately forming four risk semantic images including technological regulation,technological governance,survival adap-tation,and systemic resonance.

关 键 词：生成式人工智能 风险语义图像 技术风险 

分 类 号：C913.1[经济管理]