ETHICS

该数据集名为ETHICS，包含了五个子数据集，分别代表不同的伦理学领域：正义、义务论、美德伦理学、功利主义和常识伦理学。这些道德判断经过精心挑选，以体现人类共有的价值观。该数据集旨在通过不同的伦理框架评估人工智能对人类共同价值观的理解能力。规模上，多个子数据集包含的例子数量不一，其中正义、美德伦理学和功利主义各有100个例子，其他子数据集则有50个。该数据集的任务是评估在各种伦理场景中的道德判断。

The dataset is named ETHICS, which comprises five sub-datasets corresponding to different ethical domains: justice, deontology, virtue ethics, utilitarianism, and commonsense ethics. These moral judgments are meticulously curated to embody universally shared human values. The dataset aims to evaluate the ability of artificial intelligence to understand shared human values through diverse ethical frameworks. In terms of scale, the number of examples varies across the sub-datasets: justice, virtue ethics, and utilitarianism each contain 100 examples, while the other two sub-datasets have 50 examples respectively. The task of this dataset is to assess moral judgment across various ethical scenarios.

搜集汇总

背景与挑战

背景概述

ETHICS数据集专注于AI与人类共享价值的对齐研究，包含多个伦理子任务，用于评估和比较不同AI模型在伦理决策上的性能。数据集提供了详细的测试结果和排行榜，支持相关研究的开展。