AI Cracks Open Ancient Scroll Without Lifting a Finger

Unlocking the Past Without Touching It

An international team of researchers has successfully deciphered a 2,000-year-old scroll destroyed in the eruption of Mount Vesuvius, all without physically opening it. Using advanced imaging and machine learning, the team virtually unrolled the delicate, carbonized papyrus to reveal Greek letters and words hidden for centuries. This marks a milestone in both archaeology and AI, offering a novel method to study ancient texts that are too fragile to handle. The breakthrough came as part of the Vesuvius Challenge, a global competition aimed at decoding unreadable scrolls preserved in the Herculaneum Library. The decoded text likely comes from a previously unknown philosophical work, shedding light on ancient thought and culture.

AI’s Role in Resurrecting Lost Knowledge

At the heart of this achievement was a machine learning model trained to detect ink traces nearly invisible to the human eye. With X-ray-based CT scanning, the AI reconstructed the internal structure of the tightly wound scroll and mapped out hidden characters. Participants in the Vesuvius Challenge submitted algorithms capable of spotting subtle patterns of ink amid warped layers of papyrus. The top submission succeeded in reading hundreds of Greek letters, offering a vivid demonstration of how AI can bridge the gap between modern tech and ancient history. This innovation could pave the way for unlocking dozens more scrolls still buried and unread for millennia.