"Artificial Intelligence can be a tool for linguistic preservation as much as for economic efficiency." Discuss with reference to BHASHINI's initiatives for low-resource Indian languages.

Artificial Intelligence is usually valued for productivity. In a multilingual country, it also decides which languages survive digitally. BHASHINI is run by the Digital India BHASHINI Division (DIBD) under MeitY and covers 36 Indian text and 23 voice languages [1]. Its record shows that AI can serve both goals, but only when the models reach users.

AI for economic efficiency and governance

• Low-cost scale: BHASHINI offers translation, speech recognition, speech synthesis and transliteration as shared public services [1].
• Mass events: at Maha Kumbh 2025, it translated announcements and safety guidelines into 11 Indian languages [2].
• Institutional productivity: under the Sansad Bhashini MoU, Parliament's own records are used to build debate translation, a procedural chatbot and speech-to-text tools [3].

AI for linguistic preservation

• Sourashtra TTS: DIBD and IISc Bengaluru built four VITS models for Sourashtra, an Indo-Aryan minority language of Tamil Nadu with about 2.48 lakh speakers. The models have male and female voices in both Tamil and Sourashtra scripts [1].
• Open archive: the 28+ hour SPICOR corpus on AIKOSH is a lasting record of the language and a base for further research [1].
• Low-resource design: the models read text character by character, so they do not need pronunciation dictionaries, which small languages lack [1].
• Cultural continuity: the project supports a migrant community whose Gujarat roots were celebrated at the Saurashtra-Tamil Sangamam 2023 [5]. This upholds the Article 29(1) right to conserve a distinct language or script.

Why the two goals are not yet equal

• Data gap: Parliament already has large records [3]. Sourashtra speech had to be recorded in a studio from just two voice artists reading prepared sentences [1].
• One-way technology: only text-to-speech was released, with no speech recognition. Users with low literacy still cannot speak to services [1].
• No user yet: the release does not name any department that will use the models [1].
• Irreversible stakes: in UNESCO's words, "when we lose a language, a community loses its unique vision" [4].

In short, the AI tools that cut translation costs for Parliament and the Kumbh can also keep a small language alive. The way forward is to record more varied speakers, add speech recognition, partner with Tamil Nadu departments and publish usage data from AIKOSH. Aligned with Article 350A and the UN International Decade of Indigenous Languages (2022–32) [4], AI can serve efficiency and preservation equally.