The scroll got buried during the eruption of Mount Vesuvius in 79 A.D.; and after 2,000 years, three students made significant contributions in deciphering its text through AI tools, winning $700,000 in prize money for their work.
Artificial Intelligence (AI) technology has been more of a reality than a part of science fiction in this day and age. Like many emerging technologies, it comes with its fair share of pitfalls and dangers, especially in regard to intellectual property and the very livelihoods of people. That said, it possesses the potential to be a helpful tool that will hopefully aid its users and members of society, rather than pose a serious ethical threat. An example of this is how machine learning tools have allowed researchers to study ancient relics like never before. Such is the case for one particular scroll, which got buried and preserved during the eruption of Mount Vesuvius.
Unlocking Ancient Texts
In 2023, computer scientist Brent Seales and Github founder Nat Friedman launched the Vesuvius Challenge, reports CBS News. The called on people from around the world to build tools that would help decipher the Herculaneum script on the ancient scroll, which is one of four documents that Seales and his research team performed high-resolution computed tomography (CT) scans on. These relics come from the only surviving library of the Vesuvius disaster, a collection of around 1,000 scrolls that people unearthed from a luxury Roman villa in Herculaneum, Italy during 1750, according to Will Henshall of Time. Claudio Lavanga of NBC News adds that Julius Caesar’s father-in-law, Lucius Calpurnius Piso Caesoninus, was the likely owner of the abode.
Now, three students have won the Vesuvius Challenge’s 2023 Grand Prize, having helped decipher five percent of one of four scanned scrolls, writes Jo Marchant for Scientific American. This marks a significant milestone in the landscape of decoding ancient documents that would have otherwise gotten destroyed or remained unopened.
Trailblazing Minds
According to an official announcement from the competition’s organizers, the three individuals were not only able to meet the initial criteria they set up, but also exceeded it spectacularly. This includes unearthing “4 passages of 140 characters each, with at least 85% of characters recoverable,” with the students adding another 11 columns of text to their findings, resulting in 2,000 words total. The Vesuvius Challenges’ team of papyrologists, working “independently,” were able to recover the most amount of text from the students’ submission than all the others.
The brilliant team behind the breakthrough consists of Youssef Nader (an Egyptian PhD student in Berlin), Luke Farritor (a 21-year-old college student from Nebraska), and Julian Schilliger (a Swiss robotics student). Together, their individual contributions to the competition made it possible for experts to decipher and translate the scroll’s contents.
It was Nader’s tools that were able to read columns of text early into the competition (which won him the First Letters Prize), while Farritor is the first person in history to read an entire word from the text (“ΠΟΡΦΥΡΑϹ,” which translates to “purple”). Lastly, Schilliger’s groundbreaking work on Segmentation Tooling and the Volume Cartographer—a program that 3D-maps papyrus areas—further helped in deciphering more text from the scroll, according to the competition organizers.
The three young men will be sharing a prize of $700,000 for their work, a part of the $1 million worth of prizes from the competition’s generous donors (with runner-up teams receiving $50,000 each).
The Challenges of Deciphering
The fragile state of the scrolls is what makes the students’ breakthrough all the more impressive and meaningful. Before organizing the Vesuvius Challenge with Nat Friedman, Brent Seales and his team spent 20 years trying to decipher the scrolls’ text, according to Jo Marchant of the Scientific American. Other researchers have attempted to do the same, but not without unfortunate results.
The scrolls were preserved as compact, log-like forms due to the eruption, making their paper incredibly brittle. During attempts to physically unroll these scrolls, a fair amount of the documents broke into fragments that papyrologists struggled to put together, according to Vesuvius Challenge organizers. Fortunately, Seales was able to create a way of capturing the scrolls and their contents by scanning them with a Diamond Light Source particle accelerator near Oxford.
However, another issue arose that Seales and his team needed to address. Though they were able to produce 3D scans of the documents, actually mapping them and deciphering their inked texts would be a time-consuming challenge—at least, without the proper machine-learning technology, which the three students were able to apply.
A Collaboration of Man and Technology
Since the carbon-based ink on the scrolls were of the same density as their paper material, it was difficult for researchers to distinguish them. However, AI-based ink detection managed to do just that, separating the text and making it clearer. Meanwhile, Schilliger’s segmentation tools were able to dissect and lay out sections of the scroll so researchers could better examine their text. One can think of it as a “virtual” way of unraveling the scroll without damaging it.
Of course, as with any legitimate research endeavor, actually deciphering what the text says (and how the pieces all come together) is still work that the Vesuvius Challenge’s team of seasoned experts—which consists of papyrologists and classicists—had to handle.
Contents of the Scroll
As for the contents of the scroll, professors like Richard Janko (who teaches Classical Studies in the University of Michigan) and Robert Fowler (who teaches Greek at the University of Bristol) believe that the author is likely philosopher and poet Philodemos, reports Will Henshall of Time. The figure was a follower of the influential Greek thinker Epicurus, which is why the few translations from the ancient document contain musings on Epicurean philosophy—a way of thinking that involves the study of pleasures like food, drink, and the arts like music.
Organizer Nat Friedman elaborates on this in an X post announcing the research findings, stating that Philodemus “writes here about music, food, and how to enjoy life’s pleasures.” Friedman adds that the philosopher contests “ideological adversaries” in the text’s closing section (possibly the Stoics), stating that they “have nothing to say about pleasure, either in general or in particular.”
The competition organizers have stated that papyrologists produced “preliminary transcriptions” of the small portion of text from the scroll. The Vesuvius Challenge website writes: “In these two snippets from two consecutive columns of the scroll, the author is concerned with whether and how the availability of goods, such as food, can affect the pleasure which they provide.”
“Do things that are available in lesser quantities afford more pleasure than those available in abundance? Our author thinks not: ‘as too in the case of food, we do not right away believe things that are scarce to be absolutely more pleasant than those which are abundant,’” the website continues. Those who are interested can find the full raw translations and theories from various experts on the Vesuvius Challenge website.
An Ever-Evolving Process
Though these recent findings are truly remarkable, they’re just the beginning of a long and hopefully fruitful quest to uncover more of the scrolls’ secrets.
After all, at present, using the segmentation tools that uncover sections of the scrolls in greater detail is still a labor-intensive, manual process. As such, the organizers hope to further develop fully-automated segmentation that would speed up the deciphering process for researchers.
“In 2023 we got from 0% to 5% of a scroll. In 2024 our goal is to go from 5% of one scroll, to 90% of all four scrolls we have scanned, and to lay the foundation to read all 800 scrolls,” writes the Vesuvian Challenge’s official website.
Banner photo from the Vesuvius Challenge website.
