Building Socio-culturally Inclusive Stereotype Resources with Community Engagement

With rapid development and deployment of generative language models in global settings, there is an urgent need to also scale our measurements of harm, not just in the number and types of harms covered, but also how well they account for local cultural contexts, including marginalized identities and the social biases experienced by them. Current evaluation paradigms are limited in their abilities to address this, as they are not representative of diverse, locally situated but global, sociocultural perspectives. Our evaluation resources must be enhanced and calibrated by including people and experiences from different cultures and societies worldwide, to prevent gross underestimations or skewed measurements of harm. In this work, we demonstrate a socio-culturally aware expansion of evaluation resources in the Indian societal context, specifically for the harm of stereotyping. We devise a community engaged effort to build a resource that contains stereotypes for axes of disparity uniquely present in India. The resultant resource increases the number of stereotypes known for and in the Indian context by over 1000 stereotypes across many unique identities. We also demonstrate the utility and effectiveness of such expanded resources for evaluations of language models. CONTENT WARNING: This paper contains examples of stereotypes that may be offensive.