Researchers build AI to read cuneiform tablets and Safaitic rock carvings

A team that includes Morris Alper, an assistant professor of computer science at the University of Miami, is building AI models to read ancient inscriptions that current AI systems cannot handle, the university reports. The project has a $300,000 grant from Schmidt Sciences through its Humanities and AI Virtual Institute, of which the University of Miami receives about $100,000.

The researchers focus on cuneiform clay tablets from the ancient Near East and Safaitic rock carvings from pre-Islamic Arabia, whose text can run in spirals. They plan models that learn from few examples and turn photos of unread inscriptions into transcripts, plus an evaluation system for cuneiform reading. The team includes researchers from Ohio State University, Tel Aviv University, Bar-Ilan University and Ariel University, and plans to release the models publicly. It is one of 19 teams that shared $10 million from the institute.