DeepMind introduces AI model to decipher ancient texts

Alphabet’s DeepMind has collaborated with the University of Venice, the University of Oxford and the Athens University of Economics and Business to build a deep neural network called Ithaca, which is able to restore missing text from ancient texts. In a paper published in Nature, DeepMind stated that Ithaca was trained using natural language processing to not only recover lost ancient text that has been damaged over time, but also identify the original location of the text and establish the date when it was made. 

The study conducted by DeepMind showed that Ithaca is 62 per cent accurate in restoring texts that are damaged and 71 per cent accurate at determining the location of a text. It was also proven that it could date texts within a range of 30 years around the correct year. In 2019, DeepMind had released Pythia, Ithaca’s predecessor, which was their first system for text restoration. 

DeepMind also partnered with Google Cloud and Google Arts & Culture to launch an interactive version of Ithaca. Welcoming further research, DeepMind has also open-sourced the code as well as the pretrained model. It also announced in its blog that it was already working on other versions of Ithaca that were trained on other ancient texts. Meanwhile, historians could avail other ancient writing systems in their studies like Akkadian, Demotic, Hebrew and Mayan. 

Normally, historians demonstrated an accuracy of 25 per cent when working on their own, but the outcome improved to 72 per cent after Ithaca.