lang icon En
Dec. 30, 2024, 4:16 p.m.
5277

Revolutionizing Ancient Text Studies: AI Deciphers Vesuvius Scrolls

Brief news summary

In October 2023, Federica Nicolardi, a papyrologist from the University of Naples, examined a papyrus fragment from the Herculaneum library, which was buried during the eruption of Mount Vesuvius in AD 79. The Vesuvius Challenge uses AI to analyze ancient texts, highlighting AI's pivotal role in interpreting writings such as Greek, Latin, and Chinese Oracle Bone Script. AI, particularly through neural networks, excels in pattern recognition and text prediction. Initially, convolutional neural networks were used for reading inscriptions in the 2010s. Now, advanced transformer models like Ithaca from Oxford University are improving the restoration and dating of ancient Greek texts. AI also aids in translating historic Korean Hanja into modern Korean and addresses gaps in Mycenaean Linear B texts, working towards deciphering Linear A. The Herculaneum project includes methods like Brent Seales' "virtual unwrapping." In March 2023, the Vesuvius Challenge uncovered unknown Greek philosophical text. Despite AI's progress, human collaboration remains crucial for accuracy. As AI unveils more texts, researchers expect groundbreaking insights into ancient literature, revealing "invisible libraries" within artifacts and manuscripts, significantly enhancing our historical and cultural knowledge.

In October 2023, Federica Nicolardi, a papyrologist at the University of Naples, received an image of a fragment from a papyrus scroll burned in the eruption of Mount Vesuvius in AD 79. This scroll, one of many found in an ancient villa in Herculaneum, has been mostly unreadable due to its fragile, carbonized state. Nicolardi is now part of the Vesuvius Challenge, a project using artificial intelligence (AI) to uncover historical texts. Thanks to AI, images revealed Greek letters from the scrolls, making once-inaccessible texts legible, marking a transformative moment for her field. AI's application in this context signifies a broader shift in how ancient texts are studied. Neural networks, particularly convolutional and recurrent networks, have made it possible to fill in gaps in historical texts and translate languages, from Greek inscriptions to Chinese Oracle Bone Script.

Projects such as those by Thea Sommerschield and Yannis Assael at the University of Oxford have utilized machine learning to restore missing parts of Greek texts with impressive accuracy. Their model, Ithaca, has improved upon these methods by using transformer models to better decipher languages. In South Korea, researchers are tackling vast historical archives with AI to translate the Hanja language, while in Greece, models assist in decoding Linear B texts with potential applications for the undeciphered Linear A script. The Vesuvius Challenge itself has overcome the problem of reading text in scrolls that cannot be opened, using high-resolution CT scans and advanced AI models to virtually unwrap and read the text. Brent Seales and his team at the University of Kentucky, in partnership with the challenge, have made significant progress. Their efforts were boosted by a public contest, resulting in a readable text of Greek philosophy thought to be by Philodemus. Efforts are ongoing to refine these techniques, solve the challenge of virtual unwrapping, and apply this method to other hidden texts, with the hope of eventually reading all Herculaneum scrolls, potentially uncovering the largest discovery of texts from the ancient world.


Watch video about

Revolutionizing Ancient Text Studies: AI Deciphers Vesuvius Scrolls

Try our premium solution and start getting clients — at no cost to you

I'm your Content Creator.
Let’s make a post or video and publish it on any social media — ready?

Language

Hot news

Dec. 23, 2025, 5:21 a.m.

Career Change in 2026? The ‘easiest’ AI jobs to e…

Photo by Paulina Ochoa, Digital Journal As many pursue careers leveraging AI technology, how accessible are these roles? A new study by digital learning platform EIT Campus identifies the easiest AI jobs to enter in Europe by 2026, showing some positions require only 3-6 months of training without needing a computer science degree

Dec. 23, 2025, 5:20 a.m.

AI in Video Games: Enhancing Realism and Player E…

The gaming industry is rapidly transforming through the integration of artificial intelligence (AI) technologies, fundamentally changing how games are developed and experienced by players.

Dec. 23, 2025, 5:15 a.m.

Google's parent buys data center energy specialis…

Alphabet Inc., Google’s parent company, announced an agreement to acquire Intersect, a data center energy solutions firm, for $4.75 billion.

Dec. 23, 2025, 5:13 a.m.

AI SEO Myths Debunked: Separating Fact from Ficti…

Artificial Intelligence (AI) has increasingly become a vital tool within Search Engine Optimization (SEO), transforming how marketers handle content creation, keyword research, and user engagement strategies.

Dec. 23, 2025, 5:12 a.m.

Virgin Voyages Deploys AI Marketing Tools for Tra…

Virgin Voyages has teamed up with Canva to become the first major cruise line to implement AI-powered marketing tools on a large scale for its travel advisor network.

Dec. 22, 2025, 1:22 p.m.

AIMM: AI-Driven Framework for Detecting Social-Me…

AIMM: An Innovative AI-Driven Framework to Detect Social-Media-Influenced Stock Market Manipulation In today's fast-changing stock trading environment, social media has emerged as a key force shaping market dynamics

Dec. 22, 2025, 1:16 p.m.

Exclusive: Filevine Acquires Pincites, AI-Powered…

Legal technology firm Filevine has acquired Pincites, an AI-driven contract redlining company, enhancing its footprint in corporate and transactional law and advancing its AI-focused strategy.

All news

AI Company

Launch your AI-powered team to automate Marketing, Sales & Growth

and get clients on autopilot — from social media and search engines. No ads needed

Begin getting your first leads today