Intersectional Bias in Japanese Large Language Models from a Contextualized Perspective
This addresses bias in AI for Japanese language users, but is incremental as it extends existing bias research to intersectionality in a specific domain.
The study tackled the problem of intersectional bias in Japanese large language models by constructing the inter-JBBQ benchmark for evaluation, finding that biased output varies with context even when social attributes are equally combined.
An increasing number of studies have examined the social bias of rapidly developed large language models (LLMs). Although most of these studies have focused on bias occurring in a single social attribute, research in social science has shown that social bias often occurs in the form of intersectionality -- the constitutive and contextualized perspective on bias aroused by social attributes. In this study, we construct the Japanese benchmark inter-JBBQ, designed to evaluate the intersectional bias in LLMs on the question-answering setting. Using inter-JBBQ to analyze GPT-4o and Swallow, we find that biased output varies according to its contexts even with the equal combination of social attributes.