A new AI software developed by a team from Martin Luther University Halle-Wittenberg (MLU), Johannes Gutenberg University Mainz, and Mainz University of Applied Sciences, can now read hard-to-read cuneiform tablets. Instead of using photos, the software uses 3D models, which provide more accurate results compared to previous methods. This allows for easier comparison of the contents of multiple tablets and opens up new research opportunities.
The researchers used 3D models of nearly 2,000 cuneiform tablets, including around 50 from MLU’s collection. It’s estimated that there are around one million such tablets still in existence worldwide, many of which are over 5,000 years old. These tablets cover a wide range of topics and provide insight into humanity’s past. However, they are heavily weathered and challenging to decipher.
The cuneiform tablets are made of clay, making them difficult to read. The writing system used was complex and encompassed several languages. Because of this, optimal lighting conditions and background knowledge are required to recognize the symbols correctly, and accessing the content of many tablets at once has been a challenge.
To address this, the researchers developed an AI system based on 3D models, which deciphers characters more accurately than previous methods. This system works similar to OCR software, converting images of writing into machine-readable text. The AI was trained using 3D scans and additional data, and it is able to reliably recognize symbols on the tablets.
Although the system currently only works with two languages, there are twelve cuneiform languages in existence. In the future, the software could be used to decipher weathered inscriptions. This groundbreaking research provides new access to cuneiform tablets and opens up many new possibilities for inquiry.