DeepMind’s new technology Ithaca, has unfolded an AI-grounded technique to decipher the missing text of damaged inscriptions, identify their original location and establish the date they were created. Modern dating techniques such as radiocarbon dating cannot be used in inscriptions making its interpretation difficult and time-consuming. Many of the surviving inscriptions had either been damaged or relocated which makes its study complicated. 

DeepMind collaborated with the Department of Humanities of the Ca’ Foscari University of Venice, the Classic Faculty of the University of Oxford and the Department of Informatics of the Athens University of Economics and Business to explore how Machine Learning can be used by historians to derive conclusions from these inscriptions. DeepMind as well intend to unlock the potential for corporation between AI and historians. 

Ithaca is a Deep Neural Network, that extends the company’s previous system Pythia which focused on textual restoration. With the introduction of this new-age technology, the world has witnessed the collaboration of DeepMind with Google Cloud and Google Arts & Culture to launch a free interactive version. This partnership between the tech tycoons will be beneficial for researchers, educators, museum staff and others. To help further research in the field, they have also open-sourced their code, the pre-trained model and an interactive collaborators book. To maximize the research potential of the tool, it is embedded with visual aids such as Restoration Hypotheses, Geographical Attribution, Chronological distribution and Saliency Maps. 

The test run by the company regards that Ithaca achieves 62% accuracy in restoring damaged texts, 71% accuracy in identifying their original location and can date texts to within 30 years of their ground-truth date ranges. According to DeepMind, “The expert historians we worked with achieved 25% accuracy when working alone to restore ancient texts. But, when using Ithaca, their performance increases to 72%, surpassing the model’s performance and showing the potential for human-machine cooperation to advance historical interpretation, establish relative dating for historical events, and even contribute to current methodological debates”. They also stated that models like the Ithaca would impact the way we study and write about the most significant periods in human history.