It’s all Greek to DeepMind: AI helps historians unravel ancient texts

DeepMind has joined forces with classic scholars to develop a new AI tool that helps historians interpret and restore the text on damaged ancient Greek inscriptions.

Named after the Greek island featured in Homer’s Odyssey, Ithaca builds on an earlier text restoration system Pythia, released in 2019.

DeepMind claims the new system is also capable of identifying a text’s location of origin and the date of its creation within a 30 year period –  according to a new paper that the Google owned British-based AI lab has published with science journal Nature.

The daunting and time-consuming task of interpreting incomplete manuscripts normally falls to specially-trained epigraphists who look for textual and contextual parallels in similar inscriptions.

However, DeepMind’s research scientists say it can be challenging for a human to harness all existing, relevant data and to discover underlying patterns – which is where machine learning can be used as a tool to help historians.

Ithaca is trained on a dataset of almost 80,000 ancient Greek inscriptions, each of which is labelled with metadata describing where and when it was written.

Ithaca looks for patterns in this information, encoding it in complex mathematical models, and uses these inferences to suggest text, date and origins.

According DeepMind Ithaca achieves 62% accuracy in restoring damaged texts, 71% accuracy in identifying their original location and is able to date texts to within 30 years of their ground-truth date ranges.

While these results are promising their less than 90% accuracy rate makes the system a complementary rather than a standalone tool to be used alongside experts. To this end,  the system is already proving effective, with DeepMind claiming that historians have already used the tool to re-evaluate significant periods in Greek history.

DeepMind is making its code open source and an interactive version of Ithaca is freely available.