Workflow · Production

Students use AI models and GitHub Copilot to decode 2,000-year-old Herculaneum scrolls

The problem

The Herculaneum Papyri are too fragile to open due to carbonization, and the ancient ink did not show up in standard X-ray and tomography scans, making any reading of the texts impossible without an AI-based approach.

First attempt

Scientists had applied virtual unwrapping with tomography and X-rays—a technique that succeeded on the Dead Sea Scrolls—but the same approach failed on the Herculaneum texts because the ink remained invisible in scans.

Workflow diagram · grounded in source
1
Scroll scans captured
trigger
“Once the scientists captured these scans of the Herculaneum papyri, technologists from around the world set out to analyze them”
2
AI ink detection models
ai_action
“Youssef had created AI detection models”
3
Segmentation automation
ai_action
“Julian who made a breakthrough automating the segmentation working extensively with GitHub Copilot”
4
Copilot-assisted pipeline coding
ai_action
“I'd write one piece of code that I needed to use to achieve the next goal and if I knew what I wanted to write, I could use the auto completion tool to help me write faster. It was a huge time-saver!”
5
Letter tracing and identification
human_review
“It required tracing the writing of this ancient writer on the scroll and finding smart ways of figuring out what letter it would be based on ink deposits”
6
Code published on GitHub
output
“Housing our code on GitHub was the only thing that made sense so the community can continue to build and to have easy access to collaborate and push progress forward”
Reported outcome

The team won the Vesuvius Challenge grand prize of $700,000 and decoded portions of the 2,000-year-old scrolls, with Julian's automated segmentation software covering around 1600 cm^2 of scroll surface.

Reported metrics
Prize money won$700,000
Scroll surface segmentedaround 1600 cm^2
code writing speed with Copilothuge time-saver
Reported stack
GitHub CopilotVisual Studio CodeGitHubDiscord
Source
https://github.blog/ai-and-ml/machine-learning/how-students-teamed-up-to-decode-2000-year-old-texts-using-ai/
Read source ↗

Frequently asked questions

What did this team achieve with this AI workflow?

The team won the Vesuvius Challenge grand prize of $700,000 and decoded portions of the 2,000-year-old scrolls, with Julian's automated segmentation software covering around 1600 cm^2 of scroll surface.

What tools did this team use?

GitHub Copilot, Visual Studio Code, GitHub, Discord.

What results were reported?

Prize money won: $700,000; Scroll surface segmented: around 1600 cm^2; code writing speed with Copilot: huge time-saver (source-reported, not independently verified).

What failed first in this deployment?

Scientists had applied virtual unwrapping with tomography and X-rays—a technique that succeeded on the Dead Sea Scrolls—but the same approach failed on the Herculaneum texts because the ink remained invisible in scans.

How is this workflow AI workflow structured?

Scroll scans captured → AI ink detection models → Segmentation automation → Copilot-assisted pipeline coding → Letter tracing and identification → Code published on GitHub.