How AI is unlocking ancient texts — and could rewrite history

Artificial neural networks are being used to decipher ancient texts, from the classical stalwarts of Greek and Latin to China ’s Oracle Bone Script, ancient divination texts written on cattle bones and turtle shells.

They are making sense of archives too vast for humans to read, filling in missing and unreadable characters and decoding rare and lost languages of which hardly any traces survive.

Machine-learning models are tackling ancient languages for which only a small amount of text survives.

Ithaca restored artificially produced gaps in ancient texts with 62% accuracy, compared with 25% for human experts.

South Korean researchers are facing very different challenges as they tackle one of the world’s largest historical archives.

In tests with artificially produced gaps, the model’s top ten predictions included the correct answer 72% of the time, and in real-world cases it often matched the suggestions of human specialists.

To improve the results further, Papavassileiou hopes to add in visual data, such as traces of incomplete letters, rather than just relying on the transliterated text.