Scholarlink

<p>Gender bias in AI systems gained increasing attention, yet insights from psychology and social science are often overlooked. This review bridges AI fairness efforts with psychological principles, addressing two key questions: Which social stereotypes are most relevant for evaluating gender bias in datasets and models? And how can experimental psychology enhance the design and interpretation of bias tests? By incorporating multiple stereotypes, this review offers deeper insights into bias evaluation and highlights the importance of psychological experimental principles in study design. It explores prompt task design, philosophical foundations, and interpretations of gender bias across tasks. Emphasizing interdisciplinary collaboration, this review advocates for integrating psychological frameworks to develop more inclusive and interpretable AI systems.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Towards fairer AI: Integrating psychological insights into gender bias testing

  • 期刊论文

  • 发表日期：

    2025年9月24日

查看全文
  • Ruoxuan Li

摘要

Gender bias in AI systems gained increasing attention, yet insights from psychology and social science are often overlooked. This review bridges AI fairness efforts with psychological principles, addressing two key questions: Which social stereotypes are most relevant for evaluating gender bias in datasets and models? And how can experimental psychology enhance the design and interpretation of bias tests? By incorporating multiple stereotypes, this review offers deeper insights into bias evaluation and highlights the importance of psychological experimental principles in study design. It explores prompt task design, philosophical foundations, and interpretations of gender bias across tasks. Emphasizing interdisciplinary collaboration, this review advocates for integrating psychological frameworks to develop more inclusive and interpretable AI systems.