SAFARI: A Community-Engaged Approach and Dataset of Stereotype Resources in the Sub-Saharan African Context

2026-02-25

AI Summary

This study addresses the critical gap in generative AI safety evaluation due to the absence of stereotype-related data from sub-Saharan Africa and the resulting lack of global representativeness. To remedy this, the authors propose a culturally sensitive, reproducible, community-engaged methodology that employs multilingual telephone interviews across Ghana, Kenya, Nigeria, and South Africa. By integrating socioculturally contextualized sampling with cross-ethnic balancing, the research systematically collects expressions of stereotypes in 15 indigenous languages alongside English. The resulting high-quality dataset comprises 3,206 utterances in local languages and 3,534 in English, offering the first comprehensive, multilingual, and multiethnic coverage of stereotypes in the region. This approach transcends the conventional English- and text-centric data collection paradigms, substantially filling a key resource gap in AI safety assessment within African contexts.

Technology Category

• Natural Language Processing: Safety and Robustness
• Philosophy and Ethics of AI: Safety, Robustness & Trustworthiness
• Machine Learning: Ethics, Bias, and Fairness

Application Category

• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search
• Economics, Online Markets and Human Computation: Trust and reliance of crowd workers and data experts on GenAI
• Security and Privacy: Large-scale security measurements

Abstract

Stereotype repositories are critical to assess generative AI model safety, but currently lack adequate global coverage. It is imperative to prioritize targeted expansion, strategically addressing existing deficits, over merely increasing data volume. This work introduces a multilingual stereotype resource covering four sub-Saharan African countries that are severely underrepresented in NLP resources: Ghana, Kenya, Nigeria, and South Africa. By utilizing socioculturally-situated, community-engaged methods, including telephonic surveys moderated in native languages, we establish a reproducible methodology that is sensitive to the region's complex linguistic diversity and traditional orality. By deliberately balancing the sample across diverse ethnic and demographic backgrounds, we ensure broad coverage, resulting in a dataset of 3,534 stereotypes in English and 3,206 stereotypes across 15 native languages.

Problem

• stereotype resources
• generative AI safety
• Sub-Saharan Africa
• NLP coverage
• linguistic diversity

Innovation

• community-engaged methodology
• multilingual stereotype dataset
• Sub-Saharan Africa
• socioculturally-situated NLP
• linguistic diversity