AI Unlocks Secrets of the Herculaneum Scrolls, Revealing the Word "Purple"
In a breakthrough blending ancient history with modern technology, researchers have successfully used machine-learning algorithms to extract readable Greek text from carbonized Herculaneum scrolls. Among the deciphered words is "purple," offering a rare glimpse into the literary treasures of the Roman world.
As of the report, only 10% of the scrolls have been read.
The scrolls were discovered in a villa widely believed to have belonged to Julius Caesar's father-in-law. Preserved but carbonized by the catastrophic eruption of Mount Vesuvius in 79 AD, they have long resisted efforts to be read. Earlier attempts to mechanically unroll the scrolls often resulted in damage, leaving many still sealed and unopened.
A New Dawn for Ancient Texts
Previous methods were painstakingly slow. In the 18th century, Father Antonio Piaggio invented a device to unroll the scrolls—a process that took years. Most scrolls were simply too fragile to open mechanically. Later, modern 3D X-ray technology provided some insight, but extracting text remained a monumental challenge because the carbonized papyrus and the ancient ink are both black, making them nearly indistinguishable.
The new AI approach is changing that. Computer science professor Brent Seals emphasized the profound cultural connection felt when reading these ancient words. He noted an unexpected finding: a scroll previously cataloged as having no visible ink was found to contain ink and may be one of the oldest Roman scrolls ever discovered.
Accelerating Discovery
The speed of this new method is also remarkable. Papyrology professor Gianluca del Mastro commented on the dramatic improvement in deciphering phrases compared to the slow, manual processes of the past.
To further accelerate the work, a new $1 million prize has been announced for deciphering a complete scroll by June next year. This ambitious goal promises to unlock even more of the lost library buried by Vesuvius nearly two millennia ago.