Ancient clay tablets written in cuneiform are one of the oldest sources of information about the life of ancient civilizations. But translating these texts is a huge challenge for scholars, as it requires knowledge not only of the language but also of the cultural context. Now, however, archaeologists and computer scientists are working together to create an artificial intelligence (AI) program that can translate ancient cuneiform texts.
Cuneiform was originally developed by the inhabitants of Mesopotamia in what is now Iraq more than 5,000 years ago. Written by extruding symbols onto clay tablets, it quickly spread throughout the ancient Near East, where it was used for more than 3,000 years. Thousands of documents written in Sumerian and Akkadian using cuneiform have survived, but the translation of these texts poses a great challenge.
An AI program developed by an interdisciplinary team of scientists from Tel Aviv University and Ariel University in Israel uses a convolutional neural network to translate ancient cuneiform texts. There are two versions of the model: one translates directly from Unicode representations of the cuneiform characters, while the other requires the cuneiform to first be transliterated into the Latin alphabet.
The model that used transliterated text worked slightly better than the model that translated directly from Unicode characters. As the authors explain, each individual glyph can have one of three different functions, making the task of translation much more difficult. The program is not perfect and works best with shorter sentences of 118 characters or less.
However, the AI program is already considered a major breakthrough in the study of ancient cuneiform texts. It can help scientists study archaic languages and decipher ancient texts. In addition, the use of AI in archaeology can lead to new discoveries and insights into ancient civilizations.