AI helps read papyrus scroll burnt to crisp during Vesuvius eruption | AI (artificial intelligence) - The Guardian

2026-06-24 · Source: artifical intelligence via Google News · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Emerging Technologies & Innovation · Depth: Novice, short

Summary

Artificial intelligence has enabled the virtual unwrapping and reading of PHerc 1667, a papyrus scroll charred during the AD79 Vesuvius eruption. Researchers uncovered 20 columns of previously hidden text, spanning over a meter, without physically unrolling the fragile artifact. Dating to the second or late-third century BC, the scroll discusses stoic philosophy on ethics, art, and human behavior, with analysis suggesting the Greek philosopher Chrysippus as a possible author. This achievement stems from the Vesuvius Challenge, a global contest launched in 2023, which utilized machine-learning algorithms developed by Prof. Brent Seales. These algorithms were trained to detect ink on hidden papyrus layers from high-resolution X-ray images. The newly deciphered text explores stoic concepts like hormē (impulse) and phronēsis (practical wisdom), highlighting the role of reason in human conduct. Another virtually unwrapped scroll also revealed "Philodemus, On Gods, Book 8," confirming a multi-book work.

Key takeaway

For historians and AI scientists seeking to decipher extremely fragile ancient texts, you should prioritize AI-driven virtual unwrapping techniques. This approach, demonstrated by the Vesuvius Challenge, allows access to previously unreadable content like the Herculaneum scrolls without risking physical damage. Consider investing in high-resolution X-ray imaging and machine learning expertise to unlock new insights from other carbonized or delicate artifacts in your collections.

Key insights

AI can virtually read ancient, damaged texts without physical intervention.

Principles

• Machine learning detects subtle ink differences in X-ray images.
• Virtual unwrapping preserves fragile historical artifacts.
• Crowdsourcing challenges accelerate complex research.

Method

Machine-learning algorithms are trained to identify ink on hidden papyrus layers by discerning subtle differences in papyrus fibers within high-resolution X-ray images.

In practice

• Apply AI-driven X-ray analysis to other damaged historical documents.
• Fund or participate in crowdsourced challenges for cultural heritage.
• Explore non-invasive methods for artifact preservation and study.

Topics

• AI in Archaeology
• Herculaneum Scrolls
• Vesuvius Challenge
• Virtual Unwrapping
• Stoic Philosophy

Best for: NLP Engineer, Research Scientist, AI Scientist, Tech Journalist

Related on AIssential

• As a ‘book scientist’ I work with microscopes, imaging technologies and AI to preserve ancient texts — Advanced scientific methods preserve ancient texts, revealing hidden information and expanding access to cultural heritage.
• Unmixing ATR-{\mu}FTIR spectroscopic images of cross-sections of historical oil paintings — An unsupervised CNN autoencoder with a weighted spectral angle distance loss effectively unmixes complex ATR-μFTIR data from historical paintings.
• Accelerating science with AI and simulations — AI is at a second inflection point, integrating language and multi-modal reasoning to accelerate scientific discovery.
• Austrian Academy of Sciences - Mistral AI — Apollo, an LLM trained on 600M Greek words, revolutionizes ancient text decipherment, accelerating research from years to hours.
• Case Western Reserve Researchers Using AI at Hair-Width Scale to Reveal Renaissance Master’s Hidden Hand | Newswise - Newswise — AI and physics reveal microscopic brushstroke patterns to attribute and authenticate Renaissance art.
• The Relic Condition: When Published Scholarship Becomes Material for Its Own Replacement — Published scholarly work can be distilled into AI systems capable of expert-level academic functions.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by artifical intelligence via Google News.