程序性平权与结构性歧视:ChatGPT性别平等观的对话测试分析  

Procedural Equality and Structural Discrimination:Analysis on ChatGPT's Gender Equality from Dialogue Test

作　　者：宋素红 赵文娜 梁泽龙 SONG Suhong;ZHAO Wenna;LIANG Zelong

机构地区：[1]北京师范大学新闻传播学院,100875 [2]北京师范大学新媒体与女性研究中心,100875

出　　处：《中华女子学院学报》2024年第3期86-94,共9页Journal of China Women＇s University

基　　金：教育部人文社科基金一般项目“新媒介环境中的社会性别意识传播”的阶段性成果,项目批准号:18YJA860012。

摘　　要：ChatGPT在人机对话中是否内含了自身的性别观,成为智能传播环境中认识社会性别平等的新着眼点。通过与ChatGPT进行人机对话并参照扎根理论分析对话文本,发现它强调AI客观中立,表层认同现代性别平等价值观而深层隐含传统性别观,存在程序性平权与结构性歧视的矛盾性。生成式人工智能在深浅层次表现出不一致的性别观,得到启示:要警惕人工智能无意识的性别偏见输出,加大性别平等政策对AI技术的保障;同时构建专业大模型,以“算法先行”原则助力人工智能的强大技术力量推进社会性别平等观念的传播。The presence of gender perspective in human-computer communication within ChatGPT has become an important aspect to understand gender equality in intelligent communication environments.Through analyzing the grounded theory of dialogue text from human-computer communication with ChatGPT,it has been observed that ChatGPT emphasizes AI's objective neutrality,superficially acknowledges modern gender equality values and implicitly incorporates traditional gender views,which makes a contradictory of the procedural gender equality and structural discrimination.The generative AI displays inconsistent gender perspectives at different levels,which requires us to be cautious of the unconscious gender bias in AI's output and to strengthen the gender equality protection policies for AI technology.Meantime,we should develop professional large language models and promote the powerful technological force of AI to spread gender equality concepts with the principle of"algorithm beforehand".

关 键 词：ChatGPT 性别平等观 无意识性别偏见 算法先行 

分 类 号：C913.68[经济管理]