BengaliMoralBench is a new culturally grounded ethics benchmark for LLMs in Bengali, with scenarios in five domains evaluated zero-shot across multiple models revealing performance variation and cultural grounding weaknesses.
Mislabeling unethical prioritization of male family members illustrates systemic bias inherited from training corpora, affecting the fairness and moral alignment of predictions
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2025 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
BengaliMoralBench: A Benchmark for Auditing Moral Reasoning in Large Language Models within Bengali Language and Culture
BengaliMoralBench is a new culturally grounded ethics benchmark for LLMs in Bengali, with scenarios in five domains evaluated zero-shot across multiple models revealing performance variation and cultural grounding weaknesses.