AI for Under-Resourced Manuscript Traditions

Arabic, Coptic, Japanese, Syriac, Tibetan — researchers working with non-Latin manuscripts face a distinct set of challenges: scarce training data, few specialist annotators, and no established guidelines for many scripts and hands. This roundtable, convened by researchers from The Digital Orientalist, asks the hard questions: Are small specialist models stepping stones toward something more general, or are we building for ourselves by default? If specialist models are ever to aggregate into wider resources, what decisions about standards and documentation need to be made from the start — and who is responsible for making them? A conversation about reproducibility, inclusivity, and what a more linguistically diverse digital humanities actually requires.