Today's News

Why India’s indigenous AI models need better data for regional languages

Indian Express - Explained · Published 15 November 2026 · Last updated

India's indigenous AI models, while advancing under the IndiaAI Mission, face severe structural hurdles due to a scarcity of digital linguistic data for regional and low-resource languages.

Key points

  • On November 3, the IndiaAI Mission under the Union Ministry of Electronics and IT announced the release of three indigenous foundation language models.
  • EKA-1 was released to cover 22 scheduled Indian languages for tasks including mathematics, coding, and reasoning.
  • The models Dhīr 2B and Dhīr 5B were launched to support 10 Indian languages for information retrieval and government services.
  • These foundation models received subsidised access to national AI compute infrastructure through the IndiaAI Innovation Centre pillar.
  • The IndiaAI Innovation Centre pillar was established in 2024 to facilitate the development of indigenous large language models and domain-specific foundational models.
  • Computing power represents only one constraint, with ready and available domestic linguistic data constituting a major hurdle for regional AI development.
  • Languages such as Assamese, Bhojpuri, and Dogri suffer from a lack of digitised linguistic data and computerised text, classifying them as low-resource languages.

Background

  • The Union Ministry of Electronics and IT established the IndiaAI Innovation Centre in 2024 to build indigenous and homegrown large language models.
  • The IndiaAI Mission advanced its strategic roadmap by releasing foundation language models on November 3 to integrate regional languages into mainstream artificial intelligence adoption.

Prelims facts and Mains angle for this issue are in the full SWAदेSH note.

Original report: Indian Express - Explained

Sources