Researchers Use AI to Decipher Ancient Roman Texts Carbonized in Deadly Mount Vesuvius Eruption
A team of researchers has employed artificial intelligence (AI) to crack the code of ancient Roman texts that were burned and carbonized during the infamous eruption of Mount Vesuvius in 79 A.D. The texts, nearly 2,000 years old, were previously unreadable due to the charring they underwent in a villa located in Herculaneum, an ancient Roman town near Pompeii.
Believed to have been owned by Julius Caesar’s father-in-law, the texts suffered carbonization as a result of the intense heat generated by the volcanic debris. Although an Italian farmer unearthed the villa in the mid-eighteenth century, the ancient texts remained hidden for centuries.
Unfortunately, many of the scrolls were damaged or destroyed during early attempts to unroll them due to their delicate condition. However, the surviving scrolls were discovered to contain philosophical texts written in Greek. Nevertheless, hundreds of scrolls remain unopened and unintelligible.
A breakthrough occurred last year when Dr. Brent Seales led a team at the University of Kentucky that utilized high-resolution CT scans to successfully unroll the texts. Yet, the black carbon ink on the papyrus made it impossible to decipher the writing.
Dr. Seales and his team, in collaboration with tech investors, initiated the Vesuvius Challenge. This challenge offered a grand prize of $1 million to any team capable of retrieving four passages, each containing 140 characters, from a Herculaneum scroll.
Several individuals came together to develop an AI model that utilized pattern recognition to unravel the lettering on the scrolls. Youssef Nader, a PhD student in Berlin, SpaceX intern and student Luke Farritor, and Swiss robotics student Julian Schillinger joined forces to build the successful AI model.
Their efforts led to the successful decoding of approximately 5% of the first scroll. The passages they decoded were written by the philosopher Philodemus and discussed the concept of pleasure, which is regarded as the highest good in Epicurean philosophy. One passage in particular examines the question of whether it is easier for humans to do without things that are readily available.
With food as well, we do not immediately believe that scarce things are absolutely more pleasurable than that which is plentiful, writes the author, who contemplates whether it is more effortless for humans to forego things that are abundant. The author suggests that such inquiries should be frequently pondered.
While the deciphering of ancient texts is a significant achievement in itself, the use of AI for this purpose opens up new possibilities for unlocking hidden knowledge from the past. As more advancements are made in AI technology, it is likely that even more ancient texts will become readable, offering a deeper understanding of history and human thought.
This breakthrough serves as a testament to the power of collaboration between human intellect and cutting-edge technology, demonstrating the potential for future discoveries in the realm of historical research.
