期刊文献+

个性化推荐算法透明度对用户感知可信度的影响

Impact of Algorithmic Transparency of Personalized Recommendation on Users’Perceived Trustworthiness
下载PDF
导出
摘要 [目的/意义]人工智能技术的发展与应用在推动人类社会变革的同时,也破坏了人与技术之间的信任。探索算法透明度与用户感知可信度间的关系,有助于破除算法黑箱,重塑技术信任,发展可信赖的人工智能。[方法/过程]文章以微信朋友圈信息流广告为研究平台,招募50位被试开展眼动实验,结合问卷和访谈等方法探究在不同的透明度情境下,用户对个性化推荐广告的感知可信度,进而厘清人与算法间的信任关系,从信任视角提出透明度提升策略。[结果/结论]用户的感知可信度受到透明度程度、透明度内容和结构等因素的影响,未来应当发展合理有效、方便访问、易于理解的算法透明度,以此提高用户信任,推动可信人工智能发展。[局限]研究的样本来源和研究视角单一,后续将扩大实验规模探索不同用户群体的感知可信度差异。 [Purpose/significance]While the development and application of AI technology promotes the change of human society,it also destroys the trust between people and technology.Exploring the relationship between algorithmic transparency and userperceived trustworthiness can help break down the algorithmic black box,reshape technological trust,and develop user-trustworthy AI.[Method/process]In this study,50 participants were recruited to carry out eye-tracking experiments using WeChat moments as the research platform,combining questionnaires and interviews to explore users’perceived trustworthiness of personalized recommendation ads under different transparency scenarios,to clarify the trust relationship between people and algorithms,and to propose transparency improvement strategies from the perspective of trust.[Result/conclusion]It is found that users’perceived trustworthiness is affected by factors such as transparency level,the content and structure of transparency,etc.We should develop reasonable,efficient,accessible and comprehensible algorithmic transparency to improve users’trust and promote the development of trustworthy AI.[Limitations]This study has a single sample source and research perspective,and the subsequent experiments will be scaled up to explore the differences in perceived trustworthiness of different user groups.
作者 吴丹 武瑜轩 Wu Dan;Wu Yuxuan(School of Information Management,Wuhan University,Hubei Wuhan 430072;Center for Studies of Human Computer Interaction and User Behavior,Wuhan University,Hubei Wuhan 430072)
出处 《情报理论与实践》 CSSCI 北大核心 2024年第11期91-100,共10页 Information Studies:Theory & Application
基金 国家自然科学基金重大研究计划培育项目“人机交互视角下数据与知识双驱动的可解释智能决策方法研究”(项目编号:92370112) 2023年度湖北省自然科学基金创新群体项目“以人为本的人工智能创新应用”(项目编号:2023AFA012)的成果。
关键词 算法透明度 感知可信度 可信人工智能 个性化推荐广告 algorithmic transparency perceived trustworthiness trustworthy AI personalized advertising
  • 相关文献

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部