NVIDIA's Sewak Advocates for Culturally Diverse AI Models for India

NVIDIA executive Mohit Sewak stresses the need for multimodal AI models that consider India's cultural and linguistic diversity. Speaking at the Global AI Summit, Sewak highlights the importance of large multilingual datasets and synthetic data generation to create culturally aligned Indian LLMs. He warns against relying solely on Western-centric models.

NVIDIA's Mohit Sewak has called for the development of multimodal AI models that reflect India's rich cultural and linguistic diversity. Addressing the Global AI Summit, Sewak emphasized the necessity for large-scale, multilingual datasets and synthetic data generation to build culturally sensitive Indian Large Language Models (LLMs).

Highlighting the diverse linguistic landscape, Sewak stated, "India speaks around 10,500 unique dialects across 123 unique languages," contrasting this with existing models that handle only 100 languages. He underscored the need for models that go beyond mere translation, emphasizing voice as an essential element.

Sewak also raised concerns about the Western-centric nature of most LLMs, stressing the urgency for Bharat-specific data. According to Sewak, building iterative, hierarchically complex foundation models tailored for India could align AI development with Indian cultural sensitivities.

(This story has not been edited by Devdiscourse staff and is auto-generated from a syndicated feed.)