A deep learning artificial intelligence (AI) model has decoded ancient Mesopotamian cuneiform tablets, thanks to researchers at Israel's Hebrew University of Jerusalem (HUOJ).
The team input transcriptions of 10,000 Akkadian cuneiform tablets into the model, which was previously trained on 104 different languages.
The AI could suggest contextually accurate words and phrases to fill in the gaps in new tablets, complete tablets with known missing pieces with 89% accuracy, and expand possible textual interpretations.
HUOJ's Gabriel Stanovsky said, "The major finding we have in this work is using other languages really helped in solving Akkadian. Without pre-training the model on those 104 different languages, accuracy was nearly 30 percentage points lower."
From New Scientist
View Full Article
Abstracts Copyright © 2021 SmithBucklin, Washington, DC, USA