GeniL: A Multilingual Dataset on Generalizing Language

arXiv:2404.05866v1 Announce Type: new
Abstract: LLMs are increasingly transforming our digital ecosystem, but they often inherit societal biases learned from their training data, for instance stereotypes associating certain attributes with specific identity groups. While whether and how these biases are mitigated may depend on the specific use cases, being able to effectively detect instances of stereotype perpetuation is a crucial first step. Current methods to assess presence of stereotypes in generated language rely on simple template or co-occurrence based measures, …

abstract arxiv biases cases cs.cl data dataset digital digital ecosystem ecosystem identity instance language llms multilingual stereotypes training training data type use cases