Scholarlink

As generative artificial intelligence (AI) has gained more impact at an alarmingly increasing speed and coverage all over the world, research has raced to consolidate and identify the most important and commonly accepted ethical principles that governments, companies and institutions should employ. This article aims to serve this common purpose by proposing an internationally recognized version of AI ethical principles, and the common thread and spirit behind based on 870 articles identified in Scopus database using a text mining method. We hope to contribute to existing literature by conceptualising these AI ethical principles into a standard that could be unitarily interpreted, understood, and exercised. Based on the seven most recognised AI ethics principles, we conduct a critical content analysis on selected kinds of literature from the social science field which proposes, compares, converges, or diverges different AI ethics. Results indicated that: (1) diverse approaches are applied to derive the ethical principles across disciplines and regions/countries, (2) customization is necessary to fit particular needs for different disciplines and functions of AI technologies, (3) more empirical justification is essential, and future studies should examine the moral reasonings of people in case studies that address ethical grey areas. Some recommendations such as sunshine tests and the need for AI literacy education are proposed in today’s GPT era.

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

In Search of an International Standardised Code of Ethics on the Use of Generative AI: A Social Science Perspective

  • 章节

  • 发表日期：

    2025年

查看全文
  • Sau Wai Law,
  • Davy Tsz Kit Ng,
  • Josephine Lau

摘要

As generative artificial intelligence (AI) has gained more impact at an alarmingly increasing speed and coverage all over the world, research has raced to consolidate and identify the most important and commonly accepted ethical principles that governments, companies and institutions should employ. This article aims to serve this common purpose by proposing an internationally recognized version of AI ethical principles, and the common thread and spirit behind based on 870 articles identified in Scopus database using a text mining method. We hope to contribute to existing literature by conceptualising these AI ethical principles into a standard that could be unitarily interpreted, understood, and exercised. Based on the seven most recognised AI ethics principles, we conduct a critical content analysis on selected kinds of literature from the social science field which proposes, compares, converges, or diverges different AI ethics. Results indicated that: (1) diverse approaches are applied to derive the ethical principles across disciplines and regions/countries, (2) customization is necessary to fit particular needs for different disciplines and functions of AI technologies, (3) more empirical justification is essential, and future studies should examine the moral reasonings of people in case studies that address ethical grey areas. Some recommendations such as sunshine tests and the need for AI literacy education are proposed in today’s GPT era.