DeepMind’s new AI model helps decipher, date, and locate ancient inscriptions

Photo by: Sepia Times/Universal Images Group via Getty Images

New machine learning techniques can help archaeologists understand the past and decipher ancient texts. The latest example is an artificial intelligence model created by DeepMind that can help restore text that is missing from ancient Greek inscriptions and offer suggestions for when the text was written.

Thea Sommerschield, a historian and machine learning expert who helped create the model, told journalists that inscriptions are direct sources of evidence.

Due to their age, these texts are often damaged, making restoration a rewarding challenge. Epigraphers look for parallels in similar inscriptions to find out when they were written.

That is where machine learning can help.

Image: DeepMind

The new software, named Ithaca, is trained on a dataset of 78,608 ancient Greek inscriptions, each of which is labeled with a description of where and when it was written. Like all machine learning systems, Ithaca looks for patterns in this information and uses complex mathematical models to suggest text, date, and origins.

The scientists who created the model say it is 62 percent accurate when restoring letters in damaged texts. The geographic origins of an inscription can be traced back to one of 84 regions of the ancient world with a 71 percent accuracy.

It is important to remember that Ithaca is not capable of operating independently of human expertise. Ithaca was designed as a tool to aid historians, rather than as a fully-automated artificial intelligence historian.

Image: DeepMind

Eleanor Dickey, a professor of classics from the University of Reading who specializes in ancient Greek and Latin sociolinguists, told The Verge that Ithaca was an exciting development that may improve our knowledge of the ancient world.

DeepMind highlighted tests that showed the model helped improve the accuracy of historians restoring missing text in ancient inscriptions from 25 percent to 72 percent. Those being tested were students. She says that the models may be accessible, but that doesn't mean they can replace the specialized academics who decipher texts.

It is not yet clear to what extent use of this tool by genuinely qualified editors would result in an improvement in the editions generally available, but it will be interesting to find out. She wanted to try the Ithaca model out for herself. Anyone can test the software and open-source code online.

Ithaca and Pythia have been used to help with recent archaeological debates, including helping date inscriptions discovered in the Acropolis of Athens. The true potential of the software has not been seen.

Sommerschield believes that the real value of Ithaca may be in its flexibility. It can be configured to work with any ancient script, even if it was trained on ancient Greek inscriptions.