OPINION | A language needs more than AI

Last week, Uttarakhand’s Chief Minister Pushkar Singh Dhami recorded a voice clip in Kumaoni at his camp office under the “Bhasha AI” initiative, a voice-collection portal developed by the AiSE Foundation for Garhwali, Kumaoni and Jaunsari - the three regional languages of the state. The initiative’s long-term objective is to enable AI-based technology to understand and communicate in these regional languages, and in doing so to carry them into the digital domain and preserve them for future generations.

‘Bhasha AI’ is part of a larger national turn towards AI-based technology for linguistic preservation. In August this year, the researchers at IISc’s SPIRE Lab, in collaboration with ARTPARK and Google, have released SraVaani, the first multilingual Indian speech recognition model trained on 65 Indian languages and dialects, including 20 scheduled languages and 45 regional languages and dialects. Similarly, last year, the ministry of tribal affairs launched Adi Vaani, India’s first AI-powered translation platform for Adivasi languages, which will support digitisation of tribal tongues including Gondi, Bhili, Mundari, Santhali, among others.

Alongside these recent initiatives, efforts like BHASHINI, AI4Bharat, Gyan Bharatam Mission demonstrate India’s commitment in building a strong architecture for preservation of its linguistic and cultural heritage for its present and, most importantly, for its future generations.

Yet this optimism rests on an unexamined assumption that preserving a language is simply a matter of preserving its data. We must understand that a language is more than a static repository of words. It truly breathes when in use through daily conversations, thought, argument, humour. By that measure, even the formidable capacities of AI fall short. AI tools can store words and construct sentences, yet it cannot replicate the active exchange of words between human minds. Unless a person is speaking to another person—it is not a language. It is, at best, a keepsake of one.

Preservation, therefore, should be the beginning and not the ceiling. A language survives not simply when its words and scripts are documented, but when they travel from one generation to the next, carried through conversations at home, in classrooms, at marketplaces, and in everyday life. Without a meaningful space in socio-economic life, a language’s intergenerational transmission is hampered, and even a meticulously preserved linguistic archive risks becoming a record of a voice that society no longer speaks.

Dogri, the regional language of Jammu, offers a compelling illustration of this gap between preservation and lived reality. Included in the Eighth Schedule of the Constitution in 2003, and being covered by the AI initiatives, the language faces a troubling generational divide in Jammu. As per recent survey findings from across the 20 different regions of Jammu district, respondents aged over 60 recorded a perfect average score of 3 out of 3 in speaking, reading and writing Dogri. But the findings revealed a troubling pattern as one moves down the age ladder. Proficiency began to weaken among younger age groups, with spoken fluency declining among respondents aged 41-60. Among those under 20, reading and writing proficiency fell to zero on the survey’s scoring scale.

The findings underline a fundamental point: bringing a language into the digital world is an important step, but its future ultimately depends on whether it continues to be spoken, read and written in everyday life.

Technological intervention, therefore, must go hand in hand with community-led efforts to keep languages alive. New Zealand’s Te Reo Maori offers an important lesson in how community-led efforts can help prevent linguicide. Long before the digital initiatives like Korero Maori app, developed by Te Hiku Media, Maori communities had begun rebuilding their declining language from the ground up through the Kohanga Reo or language nests, established in 1982, where families and elders created environments for young children to learn Maori. These efforts preceded Maori’s recognition as an official language of New Zealand in 1987. This way, the Maori experience became an inspirational example for societies across the globe that preservation becomes meaningful when people feel pride in their heritage, leading to the language being passed down from one generation to another.

Beyond community efforts, to ensure that mother tongues are not confined to the domestic sphere, the state has an indispensable role to play by ensuring their place in academic institutions and in wider socio-economic life. As per the UNESCO’s ‘Bhasha Matters Report, 2025’, the preservation of India’s cultural pluralism can be ensured through a multilingual education system.

In this regard, the National Education Policy 2020 and National Curriculum Frameworks for the Foundational Stage (2022) and School Education (2023) offer a clear path forward. However, realising this potential requires addressing the institutional and pedagogical barriers to mother-tongue education, including the availability of trained teachers, institutionalising community participation, and appropriate learning materials, among others.

Therefore, a multidimensional approach—integrating multilingual education into policy-making, utilising technological tools like Artificial Intelligence, and working together with communities and stakeholders—is essential to linguistic revitalisation in India. AI alone can, no doubt, ensure that a language is not forgotten, but it’s the people who keep it alive.

(The author is a researcher and writer. He has previously been associated with a Member of Parliament. The opinions expressed in this article are those of the author and do not purport to reflect the opinions or views of THE WEEK.)