SIDDHARTH DESAI
In an age when technology is reshaping every aspect of human life, artificial intelli gence (AI) has emerged as one of the most transformative forces of our time. In India, AI is not only powering growth and driving innovation; it is also be ing harnessed to protect some thing far more enduring: our civ ilisational heritage and linguistic diversity. India’s cultural heritage is vast, rich and multi-layered. From ancient manuscripts and monumental architecture to per forming arts, crafts, folklore and indigenous knowledge systems, our heritage is living, evolving and deeply rooted in language. It is transmitted across gener ations through scripts, spoken forms, regional expressions and community memory. Each lan guage is not merely a medium of communication but a reposi tory of knowledge, identity and worldview. As of the 2011 cen sus, India recognises 22 sched uled languages, 99 non-sched uled languages, and thousands of mother tongues and tribal dialects belonging to multiple language families. Recognising that this heritage is both our inheritance and our strength, the government has launched sustained institution al, educational and digital initi atives to preserve and promote it. At the heart of these efforts is the democratisation of technol ogy — ensuring that AI serves people in the languages and for mats they understand and trust. Through emerging digital public infrastructure for languages, in cluding platforms such as Bhash ini, Anuvadini, Gyan Bharatam and Adi Vaani, India is building systems that enable translation, voice recognition, manuscript digitisation, knowledge access and the inclusion of tribal and endangered languages. The manuscript treasure in In dia is enormous. Our collections exist in more than 40 different formats — palm leaf, birch bark, cloth, paper and others — and India perhaps holds the largest corpus of manuscripts in the world. The primary challenge is decipherment. Unless these manuscripts can be read, their value is limited. Teaching com munities to read every script is an uphill task that requires the coordinated efforts of insti tutions, government agencies, universities, schools and col leges. AI, however, has become a practical aid in deciphering and transcribing these mate rials. By making manuscripts readable and searchable, AI en ables scholars to analyse their contents critically, accelerating research and widening access to our knowledge tradition. The diversity and survival of languages are also at stake. Census 2011 lists 1,369 mother tongues, languages and dialects in India. Historically, before the advent of printing, there were about 6,000 languages; roughly 4,000 of them perished because they did not adapt to printing technology. Today, with the rise of digital and AI technologies, languages that fail to adapt risk a similar fate. The National Education Policy (NEP) 2020 strongly advocates multilingual ism, encouraging every citizen to become more linguistically capable. Technology can support this goal by offering language courses remotely and enabling anyone, anywhere in the coun try, to learn and access litera ture in other Indian languages. Through translation and other AI-enabled tools, people can understand texts from distant regions, facilitating cultural ex change and strengthening na tional cohesion. AI is inherently language-ag nostic: depending on the train ing dataset, models can learn multiple languages and trans late between them. Like a child acquiring language through ex posure, AI systems learn from data and can be trained to serve many linguistic communities. This capability makes AI a pow erful medium for person-to-per son communication even when speakers do not share a common language. Tools such as Optical Character Recognition (OCR) and foundation models are al ready helping convert printed and handwritten texts into dig itised, searchable formats. En thusiasts and experts alike can contribute to building corpora from ancient Granthas and oth er manuscripts that will, in turn, support teaching, innovation, re search and preservation. Accuracy remains a key con cern. Translation platforms such as Anuvadini and Bhashini re port high accuracy for straight forward translations — often in the range of 85–90 percent. However, accuracy declines for old manuscripts, rare scripts and texts that have not been widely deciphered; in such cas es, it can fall to 65–70 percent or lower. Furthermore, many clas sical and literary texts contain meanings that are not purely lit eral; they require contextual and cultural interpretation. This is where human expertise remains indispensable. AI can accelerate and assist the work of scholars, reducing labour and speeding up workflows, but human intel ligence is essential for interpret ing nuance, recovering hidden meanings and ensuring fidelity. Another technical and cultur al challenge is script availability. Several languages lack their own standardised scripts and borrow scripts from other languages. AI and indigenous models can help by generating or standardising scripts — for example, Bharati scripts — that better reflect Indi an linguistic traditions and pro mote smoother cross-language interaction. Building indigenous AI systems that embody Indian cultural values, philosophical outlooks and grammatical con ventions will make the tech nology more relevant, inclusive and empowering. AI tools must reflect local contexts rather than imposing foreign design choices that may not capture India’s lin guistic diversity. For languages with fewer speakers, AI offers equal oppor tunity. To develop and sustain contemporary literature and to generate new vocabulary, a continuous process of word cre ation and curricular adaptation is necessary. Technology-ena bled teaching and learning can support this process, enabling entertainment, literature, educa tion and other domains to flour ish in many languages. Startups are already active in this space: more than a hundred ventures are developing AI solutions in Indian languages for purposes ranging from language learn ing to psychological counselling and student support. This en trepreneurial activity demon strates AI’s role as an enabler for democratisation—widening access to education, scientific knowledge and practical appli cations. To realise these benefits, India must ensure accessible com puting infrastructure, including GPU access for universities, cen tres of excellence focused on do main-specific AI (for agriculture, urbanisation, smart cities, etc.), and sovereign AI models that serve national priorities. Digital documentation and archiving create a permanent record that anyone can consult; when texts are digitised, they can be trans lated, taught and shared across communities. Finally, responsible use of AI is central to India’s approach. Efforts are aimed at making AI inclusive, accessible and democ ratised so that its benefits per colate to the very last person. By combining technological inno vation with scholarly expertise and cultural sensitivity, India can harness AI to preserve its linguistic and cultural heritage, strengthen social cohesion, and extend opportunities to all citi zens. (The author is an advocate by profession.)
