AI decoders have burned manuscripts from ancient times

In AD 79, Phila was buried in Herculaneum by a volcanic eruption, including her library of hundreds of papyrus scrolls. Modern experts are now using X-ray technology and artificial intelligence to read texts and make them readable.
Scientists led by computer scientist Brent Sales of the University of Kentucky revealed cross-sectional images of charred scrolls from the ancient city of Herculaneum at a press conference.

As part of the “Vesuvius Challenge” competition, Luc Varitor and Youssef Nader, two computer science students, were able to decode the hidden text. Using specially trained artificial intelligence models, they were able to recognize individual letters and even the first word in X-ray CT images of a charred manuscript from Herculaneum.

Preserved preserved manuscripts

It is important to note that compared to other manuscripts, this manuscript comes directly from antiquity. Because it has not been copied many times and passed down to our time in this way. This nearly 2,000-year-old manuscript has not been copied or edited.

As previously mentioned, the scanned scroll originally came from the library of Villa dei Papari, an ancient estate in Herculaneum, which lies at the foot of Mount Vesuvius. When the volcano erupted in 79 AD, this site was buried along with the neighboring city of Pompeii. Due to its proximity to the mountain, Herculaneum was not only buried by volcanic rock, but was also engulfed by hot mudslides. The manuscripts were charred by the ambient heat and were preserved in the volcanic mass that later turned into stone.

CT scan taken with a particle accelerator

High-resolution CT scans previously performed by Brent Sells and his team using a particle accelerator made it possible to recognize multiple letters and even an entire word in a small area, Spectrum reports. Students Luc Varietor and Youssef Nader independently decrypted the same website using artificial intelligence algorithms and arrived at the same result.