Young computer-science student uses AI to decipher Ancient Herculaneum scrolls

Luke Farritor, a 21-year-old computer-science student from the University of Nebraska-Lincoln, has made a breakthrough in deciphering the contents of ancient scrolls from the Roman city of Herculaneum using artificial intelligence (AI). Farritor developed a machine-learning algorithm that was able to read the text on a carbonized scroll, which opens up the possibility of unlocking hundreds of texts from the only intact library from Greco-Roman antiquity.

According to the Nature Journal, Farritor used his algorithm to detect Greek letters on the papyrus, including the word "πορϕυρας" (porphyras), which means 'purple.' He trained the algorithm to highlight the ink on the scroll by analyzing subtle variations in surface texture. This achievement has impressed papyrologists, such as Federica Nicolardi from the University of Naples, who described it as a dream come true.

The scrolls from Herculaneum were buried under volcanic ash during the eruption of Mount Vesuvius in AD 79, and previous attempts to open them resulted in fragmented text. Scholars believed that these scrolls could never be unrolled or read due to their fragile condition. However, Farritor's breakthrough has earned him recognition from the Vesuvius Challenge, which awards prizes for deciphering the scrolls. He received the 'first letters' prize of $40,000 for reading more than 10 characters in a small area of the papyrus.

The use of AI in the study of ancient texts has been a growing trend. Machine-learning techniques have been developed to decipher texts in various languages, aiding scholars in dating and understanding their origins. Farritor's success has raised hopes of unlocking the "invisible library," which includes texts in parchment, palimpsests, and cartonnage.

This is not the first time AI has been utilized in deciphering ancient inscriptions. In 2019, Google's AI research division, DeepMind, collaborated with the University of Oxford to analyze ancient Greek inscriptions. Their project demonstrated that AI technology outperforms humans in deciphering these fragmented and missing letters. The success of AI in deciphering ancient texts is particularly valuable as epigraphers, who traditionally attempt to decipher inscriptions, often face challenges due to missing words.

The Vesuvius Challenge, started by a group of computer scientists, aims to encourage the use of machine learning to read the Herculaneum Papyri. Farritor's achievement has boosted optimism among the challenge's founders, who labeled the $700,000 grand prize as "now definitely achievable." The application of AI in deciphering ancient texts holds great promise for uncovering the knowledge preserved in these historical documents.