Abstract

Generative Artificial Intelligence is rapidly becoming an important source of information education communication and digital services yet its benefits are not distributed equally across languages. India presents a particularly important case because millions of people interact primarily through regional languages while much of the digital and training data ecosystem remains concentrated in English and a limited number of major languages. This paper examines the social consequences of unequal representation of Indian regional languages in generative AI. It comparatively analyses language performance cultural representation knowledge accessibility digital participation and educational opportunities across Indian linguistic contexts. The study draws upon recent research on multilingual language models Indian-language benchmarks and culturally grounded AI evaluation frameworks. The findings indicate that low-resource languages face weaker model performance limited cultural representation