AI Struggles to Decode India's Multilingual Medical Manuscripts

India's *Gyan Bharatam Mission* aims to digitize millions of medical manuscripts scattered across temples, monasteries, and archives, with over 7.5 lakh already processed. However, the linguistic complexity presents formidable challenges for AI systems.

Tools like Lipikar, developed at IIT Delhi, use optical character recognition to convert scanned texts into machine-readable formats across multiple Indian scripts. Yet researchers emphasize a crucial distinction: while AI can read and search text, it cannot interpret meaning in historical and cultural context.

Ancient manuscripts complicate matters further. Scripts like Kaithi appear in multiple forms, and the same script was used for different languages while single languages appeared in various scripts. A character might have four distinct variations depending on origin and writer.

Platforms like *BHASHINI* and BharatGen support multiple Indian languages, but experts warn against treating Indian medical knowledge as isolated from broader Asian traditions. Privacy concerns also emerge, as manuscript owners resist sending sensitive materials to external servers, prompting development of localized edge-computing solutions.