Beyond Categories of Caste: Examining Caste Bias and Morality in Text-to-Image AI Models
Researchers have developed a new framework to analyze caste bias in text-to-image AI models, moving beyond simple identity categories to understand the relational aspects of caste discrimination. This approach combines algorithmic auditing with critical discourse analysis to reveal how biases are perpetuated, challenging the notion of Brahminical normativity. The work proposes an anti-caste methodology for addressing bias and fairness in AI systems. AI
IMPACT Provides a more nuanced understanding of AI bias, potentially leading to fairer AI systems.