Ancient texts offer rare, direct insight into the past but the meaning of each often remains elusive. As they are typically found in fragments and engraved on inorganic materials (like stone or metal) that evade carbon dating, they present challenges to archeologists who have struggled to decipher them, or even note when they were written. This is why DeepMind—an artificial intelligence company acquired by Google’s parent Alphabet—developed Ithaca, an AI model that restores ancient texts while offering suggestions (with a 30-year window) of when they’re from. The program draws from a dataset of around 78,000 ancient Greek inscriptions to look for patterns within them and generate text, dates and origins. While the current model is only 62% accurate when translating and 71% accurate when sourcing origin, researchers find the numbers promising as the program could open pathways to deciphering ancient texts beyond Greek. Learn more about it at The Verge.
Image courtesy of Sepia Times/Universal Images Group/Getty