How AI is decoding ancient scrolls | Julian Schilliger and Youssef Nader

Summary

TED Talks Daily: How AI is Decoding Ancient Scrolls

Episode: How AI is Decoding Ancient Scrolls | Julian Schilliger and Youssef Nader
Release Date: January 24, 2025
Host: Elise Hu

---

Introduction: Unveiling the Past with AI

In this compelling episode of TED Talks Daily, host Elise Hu introduces a groundbreaking endeavor where artificial intelligence (AI) and computer vision converge to digitally unroll and decipher ancient scrolls buried by the catastrophic eruption of Mount Vesuvius nearly 2,000 years ago. The speakers, Julian Schilliger, a digital archaeologist, and Youssef Mohammed Nader, an AI researcher, delve into their innovative approach that not only preserves historical artifacts but also unlocks lost knowledge from antiquity.

---

The Historical Enigma: Scrolls Lost to Time

Yousef Mohammed Nader sets the stage by recounting the discovery of ancient scrolls in the Villa de Papiri, Herculaneum, during the excavation following Vesuvius's eruption. These scrolls, containing the philosophical musings of Philodemus of Gadara, were mutilated by pyroclastic flows, rendering them nearly inaccessible and unreadable.

“We always think about the potential of AI changing the future. But what about the potential of AI changing the past?” – Yousef Nader [04:05]

He highlights the significance of these scrolls, which discuss themes like music and pleasure—subjects that remain relevant today—and poses critical questions about the lost knowledge they may contain.

---

Challenges of Traditional Archaeology: Destructive Unrolling

Julian Schilliger elaborates on the traditional methods of handling these fragile scrolls, where the physical unrolling process often resulted in the destruction of the artifacts. The limited technological capabilities of the past left only the most damaged scrolls intact, preserving them in their original rolled state but making their contents inaccessible.

“People were racing to excavate more of these. What knowledge is included that is not known to us now?” – Yousef Nader [04:05]

Schilliger emphasizes the need for non-destructive techniques to reveal the content without causing further damage, setting the stage for the AI-driven solution.

---

The AI Revolution: Virtual Unrolling with Computer Vision

The conversation shifts to the innovative use of CT scan technology in digitizing the scrolls, spearheaded by Brent Seals from the University of Kentucky. This technology captures detailed 3D data of the scrolls, providing a digital blueprint for virtually unrolling them.

Julian Schilliger describes the inception of the Vesuvius Challenge, initiated by Silicon Valley investor Nat Friedman to encourage global collaboration in solving the virtual unrolling problem. The challenge offered both monetary incentives and eternal glory to the team that could crack the code.

“It opened a window into the past. What's next? Let's open this window. More AI will help us access information that was so far safely locked away.” – Julian Schilliger [15:29]

Schilliger and Nader joined forces through the challenge, bringing together their expertise in digital archaeology and AI to develop a sophisticated tool that could handle the complex task of virtual segmentation and unrolling.

---

Developing the Solution: Algorithmic Innovations and Collaboration

Yousef Nader details the technical hurdles faced in creating an accurate virtual unrolling process. The primary challenges included:

1. Segmentation: Identifying and isolating individual scroll segments within the 3D CT scans.
2. Ink Visibility: Detecting the carbon-based ink on papyrus, which is not visible in standard CT scans.

Through iterative algorithm development and leveraging community feedback within the Discord platform, Nader enhanced existing tools by incorporating advanced techniques like optical flow to improve the accuracy and speed of segmentation.

“The algorithms kept improving. The main trick was you needed to prevent the models from memorizing what the previous models have learned.” – Yousef Nader [11:05]

This collaborative refinement led to a 10,000-fold improvement in the algorithm's performance, enabling the team to virtually unroll the scrolls without physical intervention.

---

Revealing the Text: Machine Learning and Pattern Recognition

The next phase involved making the invisible ink visible. Yousef Nader explains how they trained a machine learning model to detect the faint carbon-based ink by analyzing small data cubes from the CT scans. This process required the model to differentiate between ink and papyrus based on subtle variations in the data.

“The first coherent word read from an unopened papyrus sheet.” – Yousef Nader [11:05]

Through multiple iterations and by creating a robust ground truth dataset, the team successfully enabled the model to recognize and reconstruct the hidden text. Their efforts culminated in the recovery of over 2,000 letters and 14 columns of text, marking a significant milestone in digital archaeology.

---

Achievements and Future Implications

The success of the Vesuvius Challenge team not only won them the grand prize but also made headlines worldwide, showcasing the immense potential of AI in historical research. Julian Schilliger reflects on the journey:

“AI helped us in large portions, writing better code and even being part in our algorithms. It opened a window into the past.” – Julian Schilliger [15:29]

Looking forward, the team envisions expanding this technology to explore other ancient artifacts and texts, potentially rewriting parts of history by making more hidden knowledge accessible.

---

Conclusion: Bridging Past and Future with AI

In closing, Julian Schilliger emphasizes the transformative power of AI in unlocking secrets of the past:

“We do not refrain from questioning nor understanding, and may it be evident to say true things as they appear.” – Julian Schilliger [15:29]

This project exemplifies how modern technology can restore and reveal historical treasures, ensuring that the wisdom of ancient civilizations continues to enlighten contemporary society.

---

Credits:
Produced and edited by Martha Estefanos, Oliver Friedman, Brian Greene, Autumn Thompson, and Alejandra Salazar. Mixed by Christopher Faizy Bogan with additional support from EM Topner and Daniela Ballarazo.

---

Key Takeaways:

• AI and Computer Vision: Essential tools in modern archaeology for non-destructive exploration.
• Collaborative Innovation: Global challenges and open-source collaboration accelerate technological advancements.
• Preservation and Discovery: Virtual unrolling preserves ancient scrolls while uncovering lost knowledge.
• Future Applications: The success of this project paves the way for AI to explore and decode other historical artifacts.

---

Notable Quotes:

• “We always think about the potential of AI changing the future. But what about the potential of AI changing the past?” – Yousef Nader [04:05]
• “The algorithms kept improving. The main trick was you needed to prevent the models from memorizing what the previous models have learned.” – Yousef Nader [11:05]
• “AI helped us in large portions, writing better code and even being part in our algorithms. It opened a window into the past.” – Julian Schilliger [15:29]

---

This detailed exploration underscores the innovative fusion of AI and archaeology, highlighting how technology can breathe new life into ancient texts and expand our understanding of historical narratives.

Transcript

[Elise Hu]

Support for this show comes from Mint Mobile. If saving more and spending less is one of your goals for 2025, why are you still paying insane amounts of money every month for wireless? That's a question that I used to ask myself but I don't now because I switched to Mint Mobile. It's the easiest way to save. I couldn't believe how easy it was to save and I can't believe how much I was overpaying for my mobile service. You can maximize your savings with Mint Mobile with plans starting at $15 a month when you purchase a three month plan. I actually got this plan for my daughter who now has a mobile plan. She has a $15 a month three month plan and we have loved it. The coverage has been great. It's supremely reliable. So you can say bye bye to your overpriced wireless plans, jaw dropping monthly bills and unexpected overages. Ditch overpriced wireless with Mint Mobile's limited time deal and get three months of premium wireless for 15 bucks. To get this new customer offer and your new three month premium wireless plan for just 15 bucks a month, go to mintmobile.com talks that's mintmobile.com talks. Cut your wireless bill to 15 bucks a month at mintmobile.com talks $45 upfront payment required equivalent to $15 a month. New customers on first three month plan only speed slower above 40 gigabytes on unlimited plan. Additional taxes, fees and restrictions apply. See Mint Mobile for details. Foreign It's 2025 and if your B2B marketing strategy for the new year doesn't include improving your ad targeting, you know what I'm about to say, right? Your ads can get lost in the noise. LinkedIn ads can help by ensuring your message makes it to the right audience. And that is so key if you are running a business or you are trying to find the best people for the openings in your company. With LinkedIn ads, you can precisely reach professionals who are more likely to find your ad relevant. Stand out with LinkedIn ads and start converting your B2B audience into high quality leads. Today you will have direct access to and build relationships with decision makers. A billion members, 130 million decision makers. I know as a small business owner I have to be on LinkedIn because that's where everyone's at. And if you are trying to propel your B2B business in 2025, what better place to be than LinkedIn? 79% of B2B content marketers said LinkedIn produces the best results for paid media. Go where the best is start converting your B2B audience into high quality leads today. We'll even give you a $100 credit on your next campaign. Go to LinkedIn.com TedAudio to claim your credit. That's LinkedIn.com TedAudio terms and conditions apply. LinkedIn the place to Be to Be.

[Marc Maron]

Thumbtack presents the ins and outs of caring for your home. Out procrastination, putting it off, kicking the can down the road in plans and guides that make it easy to get home projects done out carpet in the bathroom like why? In knowing what to do, when to do it and who to hire. Start caring for your home with confidence. Download Thumbtack today.

[Elise Hu]

Foreign you're listening to TED Talks Daily where we bring you new ideas to spark your curiosity every day. I'm your host, Elise Hu. Today's speakers did something rather remarkable. They used advanced AI and computer vision to virtually unroll ancient scrolls buried long ago by the eruption of Mount Vesuvius. Machine learning researcher Yusuf Mohammed Netter and digital archaeologist Julian Schilliger shed light on their approach and explain how AI can open up new possibilities for understanding ancient knowledge.

[Yousef Mohammed Netter]

We always think about the potential of AI changing the future. But what about the potential of AI changing the past? My name is Yousef Nader. I'm an Egyptian AI researcher and a PhD student at the Free University in Berlin. And last year I led the Vesuvius Grand Prize winning team on exploring this very question. You see, the story starts almost 2,000 years ago. A Greek philosopher that we believe was Philodemus of Gadara, Philippines, sat in one of the many rooms of the Villa de Papiri. He talked about music, he talked about pleasure. He talked about what makes things enjoyable questions that still plague us until today. One of his scribes wrote down his thoughts on sheets of papyrus. The sheets were rolled and stowed away for later generations. Fast forward 150 years, not even more. Mount Vesuvius erupts, burying Herculaneum, the villa and the words of the philosopher under a sea of hot mud and ashes. Now fast forward again to the 17th century. People are excavating around the area. They found beautiful statues, breathtaking frescoes and some weird looking pieces of charcoal. This is when the first scrolls were discovered and people were racing to excavate more of these. What knowledge is included that is not known to us now? What things should we know about these scrolls?

[Julian Schilliger]

My name is Julian and I am a digital archaeologist. When the pyroclastic flow hits hit the scrolls, it had a destructive effect. It tore into Them shred off pieces and it charred them badly. People250something years ago were curious what's lying inside those scrolls hidden and not accessible anymore because of a lack of technology, they had to resort to to physically unrolling and thereby destroying most of the scrolls. To this day, only the most damaged and deformed scrolls remain in their initial rolled up configuration. Fast forwarding a little bit. The computer age arrives. Youssef and I are born. We are going on and getting our education and well at the same time. Brent Seals, a researcher and professor, had the idea to use CT scan technology to actually digitalize the scrolls with the hope of one day digitally unrolling them. It's a difficult question how to unroll this digitally. Nat Friedman, a Silicon Valley investor, also saw this research and he wanted to help. That was in 2022. He reached out and together with Brand Seals, they created the Vesuvius Challenge with the goal to motivate nerds all over the world to solve this problem. They created a grand prize promising eternal glory and monetary incentives to anyone who could do that. I myself, I saw that in the Internet while writing my master thesis at ETH Zurich in robotics and I was instantly happy to solve it. At least try, you know. And I went on, joined the Discord community where all the people that were also contestants and playing with the scroll data were exchanging ideas and joined there and, and started working on it Also there on Discord, I met Yousef and Luke who would become my teammates and with whom I would actually win the grand prize. Surprisingly, it went on and made global headline news. It even got into the British tabloids. So when we started, there were two main problems still remaining. One, you had to unroll the scroll and two, you then had to make the ink visible. Youssef will tell you more about that part. For me, the most exciting thing was the computer vision problem of unrolling those scrolls virtually. I decided to iterate on a tool that was created by the Kentucky researchers and make it faster, less prone to errors and just iterate on it and make it better. The Vesuvius Challenge team saw that and also implemented a team of 10 people that would use my tool. They would annotate scroll data where they created a red line where the surface would lay. The algorithm then would take it into 3D space creating a three dimensional representation of the surface. Computer algorithms would then flatten it and create a segment. This all would be called segmentation in the space of the scrolling and unrolling community. So I created open source commits to this tool and implemented new algorithms from my studies like optical flow to better track the sheets during the through the volume. First off, those were really small segments and I added improvement, made the code faster and had lots of feedback from the community. They were really happy and I was happy getting lots of feedback. It was a really positive environment. So in the end I could track the performance of the algorithms, how the segmentation team performed and I could see that my improvements from start, start to finish would be around about a 10,000 fold improvement over the initial version. This algorithm was then also used to unroll all the area that you can see in our submission. All the sheets were generated with these methods. In December, I was looking for teammates. I made a blog post and I showcased my newest algorithms reaching out to anyone that was willing to team up. Youssef and Luke got into contact with me. They were happy to team up and I was happy as well.

[Yousef Mohammed Netter]

So after the virtual unwrapping, the words still are not visible. The main problem is that the ink that was used at the time was a carbon based ink. And carbon based ink on carbon based papyrus in a CT scan isn't visible, or at least to the naked eye. So the same team at the University of Kentucky decided to test whether the ink is present at all in the CT scans. For this, they took some of the pieces that people broke off the scrolls and they fed them into the same pipeline of the X ray CT scanning. And this gives us the 3D data that we're working with. You can also, because you can see the ink and it's like an exposed surface, you can even improve it with infrared imaging. And this gives you a ground truth of what letters you're actually trying to find. And then from there you can train a machine learning model to try to find these letters. The way this works is that the model looks at very small cubes at a single time and tries to decide whether there is ink present in this area or not. And then when you keep moving this cube all around, the model gets to see different data samples and and then tries to understand what ink actually is. So this is how it looks. While the model is training, the model is starting to see the letters perfectly. So the data is there, the ink is there, but it's just very hard to find and see. There's a different way that you can find this ink, one that actually scales very, very well. So this is where my journey begins with the Vesuvius challenge. There is this neat idea in computer vision literature where if you don't actually have labels. If you don't have the goal that you want your AI model to reach, you can pick an intermediary goal along the way. So one idea to just let it know about the structures and familiarize it with the data is to tell it, show it different views of the same image and tell it that these are the same images. And after that, you take this model and you train it like the previous models that the University of Kentucky did. And while the approach doesn't fully work, it also doesn't fully not work. And the first image that was produced by the model, there was some very faint signal in there. It seemed like the model is catching on something, but it wasn't clear exactly what the model is catching on. So I decided to take this predictions and create a new ground truth. Asking the model, hey, I think these might be letters. I think there's something in there. Try to find more of this. And my ground truth actually has four correct letters and four other delusions. But that was okay. So training a new model with this data, the model started to find more ink, find more letters, and the lines even looked complete. So I thought, what are the chances that if I do this again, the models keep improving. And this was the core behind our grand prize winning solution. Repeating this process over and over, the models kept improving. The main trick was you needed to prevent the models from memorizing what the previous models have learned. You're essentially asking the model to learn what the other model has learned. So overfitting was a serious problem that required a lot of experiments, but in the end, getting the recipe right, we were able to predict all of these letters without the models ever seeing them. These were the first 10 letters. This was the first coherent word read from an unopened papyrus sheet. From there, scaling the process within weeks, we had now columns of text, even special characters that the papyrologists found very interesting that the model was able to find. The approach was open sourced and, and the data and the code was out there. And the race for the grand prize was on. Recovering four paragraphs at an 85% clarity. And the key to our success was perfecting the data and the model. With so many iterations and so many experiments, in the end we were able to recover more than 14 columns of text and 2,000 letters.

[Julian Schilliger]

2,000 characters safely stored away two millennia ago. In just nine months, we discovered them again. AI helped us in large portions, writing better code and even being part in our algorithms. It opened a window into the past. What's next? Let's open this window. More AI will help us access information that was so far safely locked away. In the words of the we do not refrain from questioning nor understanding, and may it be evident to say true things as they appear.

[Elise Hu]

That was Yousef Mohammed Netter and Julian Schilliger at ted Aiviena in 2024. If you're curious about Ted's curation, find out [e-mail] curationguidelines and that's it for today. TED Talks Daily is part of the TED Audio Collective. This episode was produced and edited by our team, Martha Estefanos, Oliver Friedman, Brian Greene, Autumn Thompson and Alejandra Salazar. It was mixed by Christopher Faizy Bogan. Additional support from EM Topner and Daniela Ballarazo. I'm Elise Hu. I'll be back tomorrow with a fresh idea for your feet. Thanks for listening.

[Julian Schilliger]

My dad works in B2B marketing. He came by my school for Career Day and said he was a big roas man. Then he told everyone how much he loved calculating his return on ad spend.

[Yousef Mohammed Netter]

My friends still laugh at me to this day.

[Elise Hu]

Not everyone gets B2B, but with LinkedIn you'll be able to reach people who do. Get $100 credit on your next ad campaign. Go to LinkedIn.com results to claim your credit. That's LinkedIn.com results. Terms and conditions apply. LinkedIn the place to Be To Be.

[Unknown]

Hey, it's Marc Maron from WTF here to let you know that this podcast is brought to you by Progressive Insurance. And I'm sure the reason you're listening to this podcast right now is because you chose it well. Choose Progressives Name your price tool and you could find insurance options that fit your budget so you can pick the best one for your situation. Who doesn't like choice? Try [e-mail] and now some legal info. Progressive Casualty Insurance Company and affiliates Price and coverage match Limited by state law not available in all states if you.

[Marc Maron]

Wear glasses, you know how hard it is to find the perfect pair. But step into a Warby Parker store and you'll see it doesn't have to be. Not only will you find a great selection of frames, you'll also meet helpful advisors and friendly optometrists. Yep, many Warby Parker locations also offer eye exams. So the next time you need glasses, sunglasses, contact lenses or a new prescription, you know where to look. To find a Warby Parker store near you. Or to book an eye exam, head over to warbyparker.com retail.