BHASHINI provides AI language services to citizens towards full 'societal inclusivity'

by IANS |

New Delhi, Jan 14 (IANS) National language AI platform BHASHINI has moved from rule-based systems to AI-powered inclusive engines, providing language services to all citizens and advancing towards full societal inclusivity, according to senior government officials.

The Digital India BHASHINI Division at IT Ministry organised ‘BHASHINI Samudaye: Strengthening India’s Language AI Ecosystem’ here in collaboration with Wadhwani AI.

According to a ministry statement, the event represented a significant step in consolidating India’s language AI ecosystem through collaborative engagement, participatory governance, and shared ownership of public digital infrastructure.

“Samudaye is about building a living ecosystem — data creators, annotators, translators, developers, users, and governments — co-developing language technology together, with shared value and responsibility,” said Amitabh Nag, CEO, Digital India BHASHINI Division (MeitY).

Professor Girish Nath Jha, School of Sanskrit and Indic Studies, Jawaharlal Nehru University, said that “BHASHINI demonstrates the importance of structured dataset creation, domain-specific accuracy, and community participation. This approach is critical for developing AI systems that reflect India’s linguistic diversity and societal needs.”

The discussion highlighted the collaborative role of government agencies, academic institutions, civil society organisations, and industry stakeholders in advancing language AI initiatives across India.

According to the Ministry, the workshop convened language experts, academic institutions, civil society organisations, and data practitioners to examine use cases and identify pathways for co-creating, governing, and scaling language AI solutions under the National Language Translation Mission (NLTM).

“The success of BHASHINI lies in its comprehensive approach to translation: covering sentence-level, discourse-level, and conversational translation across all 22 scheduled languages, including non-scheduled languages and dialects,” said Shobha L., Member Research Staff, AU-KBC Research Centre.

Addressing the challenges in Dravidian languages, which often lack Sanskrit equivalents, requires expanded datasets and continued collaboration. BHASHINI’s work ensures no language is left behind, she added.

BHASHINI, in collaboration with the Gates Foundation and implemented by Civic Data Lab, launched the Dataset Onboarding Supporting Team (DOST) to support the systematic identification, preparation, and onboarding of high-value datasets into BHASHINI and AI Kosh.