A research team led by Nagaland University has proposed a framework to evaluate the use of artificial intelligence (AI) in preserving India’s cultural heritage, including manuscripts, endangered languages, oral traditions and historical records.
Published in the peer-reviewed, open-access journal Acta Innovations, the study brings together researchers from institutions in Nagaland, Tamil Nadu, Delhi, Uttar Pradesh, Telangana and Karnataka.
The framework assesses AI-assisted preservation across nine areas: digitisation accuracy, text recognition, classification, translation, visual reconstruction, descriptive information, cultural context, accessibility and ethics. It combines technical evaluation with reviews by subject experts and relevant communities.
Prof. Ravi Ranjan Kumar of Nagaland University’s Department of Hindi said technical accuracy did not necessarily ensure cultural accuracy, as AI could transcribe documents or translate speech while missing historical references, cultural nuances and the significance of traditions. He stressed the need for expert verification and community consultation, particularly for less commonly used languages, unusual scripts, damaged documents and oral histories.
Vice-Chancellor Prof. Jagadish K. Patnaik said AI could create new opportunities to document and transmit cultural knowledge to future generations, provided technological interventions respected cultural contexts, community rights and ethical considerations.
The researchers identified five areas for AI-assisted preservation: printed and handwritten documents; photographs and artworks; oral histories and endangered languages; archaeological and architectural heritage; and folklore and community knowledge. Potential applications include searchable historical records, organised image collections, transcription and translation of oral recordings, damage identification in artworks, and digital maps and three-dimensional models of heritage sites.
The study distinguished restoration, which preserves or recovers existing material, from reconstruction, which may involve inferring missing information. It recommended retaining original digital files, documenting AI-generated changes and clearly identifying uncertainties to distinguish historical evidence from computer-generated interpretations.
Highlighting the relevance of the framework to multilingual regions such as Northeast India, the researchers called for safeguards on data use, privacy and access, alongside community participation in decisions on recording and sharing heritage. They also recommended testing the framework across collections, scripts and dialects, documenting errors, incorporating feedback and ensuring the long-term usability of digital records.
The paper was co-authored by Prof. Kumar and six researchers from institutions in Tamil Nadu, Delhi, Uttar Pradesh, Telangana and Karnataka.
The study is available under DOI: 10.62441/actainnovations.v63i1s.1506.
