Researchers from Tel Aviv University and Ariel University have developed an artificial intelligence (AI) model that can automatically translate Akkadian text written in cuneiform into English. Cuneiform is one of the oldest forms of writing known, and Assyriology experts spend many years studying it to understand ancient Mesopotamian texts.
Dr. Shai Gordin of Ariel University and Dr. Gai Gutherz, Dr. Jonathan Berant, and Dr. Omer Levy of TAU trained two versions of the AI model – one that translates Akkadian from representations of cuneiform signs in Latin script and another that translates from unicode representations of the signs.
The first version, which uses Latin transliteration, produced more satisfactory results in this study, achieving a score of 37.47 in the Best Bilingual Evaluation Understudy 4 (BLEU4) which means the model can provide output comparable to an average machine translator from one modern language to another. This is a significant achievement given that there is a cultural gap of more than 2,000 years in translating ancient Akkadian.
Their findings were published in the journal PNAS Nexus. This new technology could revolutionize the study of ancient history, making it more accessible and open to a wider audience. The same group of researchers had developed an AI model called 'the Babylonian Engine' in 2020. The contemporary model is supposedly a better and reworked version of it.
These models are proposed for training students to better comprehend ancient texts and for more competent researchers as a first-stage scanning tool.
Assyria, located in the Mesopotamian plain (modern day Iraq), was named after the god Ashur, the highest in the pantheon of Assyrian gods. In 721 BC, Assyria captured the Northern part of what is today Israel, taking the Ten Tribes into captivity. These Jewish exiles lived alongside the Assyrian people, using cunieform writing."
Historians note that hundreds of thousands of clay tablets from ancient Mesopotamia, written in cuneiform, have been found by archaeologists, far more than can be translated by the limited number of experts who can read them.