Mind Matters Natural and Artificial Intelligence News and Analysis
Volcano eruption with lava flow in dark
Image licensed via Adobe Stock

AI Decodes Scrolls Scorched by Vesuvius’ Eruption

In 79 AD, Vesuvius reduced a library to charcoal. Remarkably, machine learning technology has begun to decipher scrolls that humans could not unwrap

In 79 AD, Mount Vesuvius erupted, famously destroying the resort towns of Pompeii and Herculaneum, south of Rome. The human death toll is estimated at 12,000. But ancient scrolls perished too, including a library in Herculaneum, charred and buried in ash. When no-longer-decipherable scrolls from the library were rediscovered in 1752, they were scattered in libraries throughout the world, in case … what?

Well first, the eruption:

The town, buried in ash, began to be painstakingly excavated in the eighteenth century and work continues today. That sudden doom for so many human souls resulted, at least, in a detailed picture of life in the early Roman empire. Relative to most archeology, the masses of details make it seem more like a video than a picture. But now, what about those scrolls, reduced to charred lumps? Any ordinary human effort to decipher them would destroy them.

In recent years, a combination of AI and X-ray technology is enabling decipherment. For example, a few years ago, the En-Gedi scroll, a charred lump, was deciphered via scans and image-processing algorithms. It turned out to be fragments of the Book of Leviticus, the earliest ever found in a Holy Ark.

Some enthusiasts for the new technology decided to declare a contest, Vesuvius Challenge, to decipher the Herculaneum library, believed to have belonged to Julius Caesar’s father-in-law L. Calpurnius Piso Caesoninus:

The Vesuvius Challenge was set up by Nat Friedman, Daniel Gross, and computer scientist Brent Seales with a prize fund of over $1 million (€930,000) and a top prize of $850,000 (€790,000) to go to anyone who could successfully decoding 5% of one of the scrolls of the papyri.

Jonny Walfisz, “AI helps decode scrolls destroyed during the Vesuvius eruption,” Yahoo News, February 7, 2024

The contest closed at midnight on New Year’s Eve. The winners, Luke Farritor (US), Youssef Nader (Egypt), and Julian Schilliger (Switzerland), trained machine-learning algorithms to decipher about 2,000 characters from the end of a scroll of a previously unknown text in Greek. The scroll had previously been 3D-scanned at the University of Kentucky:

So what was the scanned scroll about?

Scholars are hard at work to fully understand the meaning of what the team discovered, but initial readings believe it may be the musings on pleasure of Epicurean scholar Philodemus, who is believed to be the philosopher-in-residence where the scrolls were found.

The end of the scroll remarks on the value of things in abundance over the value that comes from scarcity: “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.”

Margherita Bassi, “Three Students Just Deciphered the First Passages of a 2,000-Year-Old Scroll Burned in Vesuvius’ Eruption,” Smithsonian Magazine, February 6, 2024

Clearly, lots more work will be needed to decode more and find out more. But then there is more prize money too:

But despite the grand prize announcement, the Vesuvius Challenge is far from finished—the newly translated text makes up just 5 percent of a single scroll, after all. In the same X announcement, Friedman revealed the competition’s next phase: a new, $100,000 prize to the first team to retrieve at least 90 percent of the four currently scanned scrolls.

Andrew Paul, “2,000 new characters from burnt-up ancient Greek scroll deciphered with AI,” Popular Science, February 9, 2024

And, we are told, there are many more charred scrolls waiting. One thing AI has certainly done is created huge new opportunities for classical scholars — which is hardly what we would have expected from popular assumptions. But anyone who hopes to take advantage of those opportunities should begin by majoring in topics like Latin, Greek, and Hebrew, the very traditional ones, as well as computer science.

Here’s more on the technology used:

You may also wish to read: Does AI challenge Biblical archeology? Sadly, many surviving documents are so damaged that they cannot be read using traditional methods. The more scrolls are deciphered using new AI methods, the more archeologists will have to study and write about.

Mind Matters News

Breaking and noteworthy news from the exciting world of natural and artificial intelligence at MindMatters.ai.

AI Decodes Scrolls Scorched by Vesuvius’ Eruption