AI models makes precise copies of cuneiform characters

Cornell and Tel Aviv University researchers develop ProtoSnap AI model that accurately copies and deciphers cuneiform characters from ancient tablets, facilitating easier reading of these historical scripts. The AI uses diffusion models that align prototype characters with variations found on tablets, boosting recognition of characters previously challenging due to variations across eras and writers. This advancement could expedite tablet-copying and allow in-depth analysis of historical data.

• Cuneiform is one of the oldest known writing forms.
• ProtoSnap 'snaps' character prototypes onto tablet images.
• Diffusion models calculate character image similarities.
• The research aims to enable manipulation of big historical data.
• Approximately 500,000 cuneiform tablets remain untranslated.