arXiv Open Access 2025

JETHICS: Japanese Ethics Understanding Evaluation Dataset

Masashi Takeshita Rafal Rzepka
Lihat Sumber

Abstrak

In this work, we propose JETHICS, a Japanese dataset for evaluating ethics understanding of AI models. JETHICS contains 78K examples and is built by following the construction methods of the existing English ETHICS dataset. It includes four categories based normative theories and concepts from ethics and political philosophy; and one representing commonsense morality. Our evaluation experiments on non-proprietary large language models (LLMs) and on GPT-4o reveal that even GPT-4o achieves only an average score of about 0.7, while the best-performing Japanese LLM attains around 0.5, indicating a relatively large room for improvement in current LLMs.

Topik & Kata Kunci

Penulis (2)

M

Masashi Takeshita

R

Rafal Rzepka

Format Sitasi

Takeshita, M., Rzepka, R. (2025). JETHICS: Japanese Ethics Understanding Evaluation Dataset. https://arxiv.org/abs/2506.16187

Akses Cepat

Lihat di Sumber
Informasi Jurnal
Tahun Terbit
2025
Bahasa
en
Sumber Database
arXiv
Akses
Open Access ✓