JETHICS
收藏资源简介:
JETHICS是一个用于评估人工智能模型道德理解能力的日语数据集,包含约7.8万个道德示例,分为四个基于伦理和政治哲学的规范性理论和概念类别,以及一个代表常识道德的类别。该数据集的构建方法遵循了现有的英语ETHICS数据集。JETHICS旨在解决当前大型语言模型在道德理解方面的不足,为促进人工智能的道德发展提供数据支持。
JETHICS is a Japanese-language dataset designed to evaluate the moral understanding capabilities of artificial intelligence models. It contains approximately 78,000 moral examples, divided into five categories: four categories based on normative theories and concepts rooted in ethical and political philosophy, and one category representing commonsense morality. The construction methodology of JETHICS adheres to that of the existing English-language ETHICS dataset. JETHICS aims to address the current shortcomings in the moral understanding of large language models, providing data support for promoting the moral development of AI.




