Generative AI answers and LGBTQ+ bias in diverse religous and country contexts

Prior research underscores the importance of enhancing the cultural sensitivity of generative AI tools, but often overlooks the intricate challenges of managing content related to minorities, whose perceptions vary widely across different cultures and religious backgrounds. This investigation explored how two generative AI platforms addressed homophobic comments when provided with various cultural and religious background information on the hypothetical user. The results indicated that AI tools can modify their support levels and types for LGBTQ+ individuals based on the cultural or religious information provided. This research contributes to the broader understanding of the social and ethical ramifications of AI responses, advocating that efforts to enhance the cultural diversity of generative AI outputs must be firmly rooted in the principles of human rights. The data of our project can be found here.