Auditing Google Search Results for Human Rights

• In Event: AI, Governance, and Power

Abstract

Research on international norms emphasizes the role of information, but often relies on a key premise: that the messages produced by international actors reach intended audiences. Yet this assumption is rarely studied directly, leaving us with scant knowledge about what people actually learn when accessing information about norms like human rights. This research note reports findings from an audit of Google search for human rights queries across 146 countries and 70 languages. We find that search results vary markedly across countries, indicating that different users are directed to different resources about human rights. Some people receive information produced by prominent international organizations, while others learn about human rights mainly from governments and local publishers. These disparities likely reflect differences not only in the sources of content, but also in the substantive messages about human rights. Furthermore, we find that natural language—far more than any other factor—correlates strongly with variation in search results, suggesting a prominent “language barrier” to international human rights discourse. Our results have significant implications for debates surrounding norm diffusion and knowledge-production in world politics, while offering practical guidance for human rights advocates.

Author

• Rochelle Layla Terman, University of Chicago