Latest news with #UniversityofWürzburg
Yahoo
08-05-2025
- Science
- Yahoo
Researchers Decipher Herculaneum Scrolls Found at Pompeii
Researchers have virtually unfurled a Herculaneum scroll which was rescued from Pompeii after the eruption of Mount Vesuvius in 79 A.D. The Herculaneum scrolls, first discovered within Pompeii's ruins in 1752, have long beguiled scientists. Written on papyrus using carbon-based ink, the scrolls cannot be physically opened without disintegrating. With the advent of digital AI technology, researchers have been given a new opportunity to discover the scrolls' the technology, they were able to decipher the contents and even identify the author of an ancient piece entitled 'On Vices.' The tome, on how to live a virtuous life, was written by Greek Epicurean philosopher Philodemus, who lived in Herculaneum from 110 B.C. until 35 B.C. The discovery of his work within the ruins of Pompeii indicates that Philodemus' work was still highly regarded amongst modern people. 'Finding the full title of a work inside a scroll that has been unread for two millennia is an astonishing achievement,' Richard Ovenden, who works at the library which houses the scrolls, told the Daily Mail. '[It] illustrates the huge potential for AI to transform arts and humanities scholarship, our understanding of the past, and breathes new life into ancient artifacts.' Although the AI technology allows researchers to virtually unroll the scrolls, reading what was inscribed within them is still a difficult task. The document was eventually unfurled and decoded by two separate research teams: Marcel Roth and Micha Nowak from the University of Würzburg, Germany; and Sean Johnson with Vesuvius Challenge, a program which awards cash prizes to those who are able to decipher the scrolls. 'The simultaneous reproduction of the title image from multiple sources, along with independent scholarly review, provides a high degree of confidence in the reading,' the Bodleian Libraries said. As it turns out, their hard work paid off. Roth and Nowak took home the very first Vesuvius Challenge Title Prize, which included a reward of $60,000.
The Irish Sun
06-05-2025
- Science
- The Irish Sun
Title of scroll burnt to a crisp by Mt Vesuvius 2,000 years ago finally read by AI – & we even know who wrote them
RESEARCHERS have deciphered the title and author of a charred scroll destroyed by the Mount Vesuvius eruption 2,000 years ago. The tightly bound scroll cannot be opened since its delicate, burnt pages would only fall apart. Advertisement 6 The scroll was discovered in Herculaneum, an ancient Roman town that was buried under the massive volcanic fallout from Mount Vesuvius Credit: Vesuvius Challenge 6 After 'virtually unwrapping' scroll PHerc. 172, researchers discovered it was titled On Vices by the Greek philosopher Philodemus Credit: Vesuvius Challenge 6 The text offered ancient Romans guidance for cultivating a virtuous life, according to researchers Credit: Vesuvius Challenge 6 The University of Oxford's Bodleian Library holds a number of the scrolls, which have been left untouched for decades Credit: Vesuvius Challenge Therefore, researchers have used a cutting-edge technique hinging on artificial intelligence (AI) to reveal the lost text. The scroll was discovered in After 'virtually unwrapping' scroll PHerc. 172, researchers discovered it was titled On Vices by the Greek philosopher Philodemus. The text offered ancient Romans guidance for cultivating a virtuous life, according to researchers. Advertisement READ MORE ON ARCHAEOLOGY Philodemus, who lived between roughly 110 and 30 BCE, was an Epicurean philosopher and poet from Gadara. The University of Oxford's Bodleian Library holds a number of the scrolls, which have been left untouched for decades. But recent advancements in technology have allowed the scrolls to be 3D reconstructed with powerful X-ray beams without damaging them. Once researchers have identified each layer of the scroll, AI is used to detect the ink. Advertisement Most read in Tech The ink is digitally painted onto the scroll, for researchers to decipher. As part of an international competition, researchers have found the title and author inside of a sealed scroll for the first time. The Vesuvius Challenge awards money to all those who can help decipher the charred scrolls housed at the Bodleian Libraries at the University of Oxford. The Tragic History of Pompeii Researchers behind this latest feat have been awarded the Vesuvius Challenge First Title Prize, which includes prize money of $60,000. Advertisement This was achieved by two parties at roughly the same time - Sean Johnson from the Vesuvius Challenge, and Marcel Roth and Micha Nowak from the University of Würzburg. The pages of the relic PHerc. 172 were first published in February, the title page had yet to be deciphered. PHerc.172 is one of roughly 800 scrolls unearthed in the Villa of the Papyri, a luxurious estate believed to have belonged to Lucius Calpurnius Piso Caesoninus, Julius Caesar's father-in-law. How AI & X-rays are unravelling scroll secrets AI and X-rays are revolutionising the study of ancient scrolls by enabling researchers to virtually "unroll" and decipher texts that are too fragile to handle physically. X-rays, particularly advanced methods like X-ray micro-computed tomography (micro-CT), create highly detailed 3D scans of scrolls. These scans capture the layers of tightly wound or damaged parchment or papyrus without physically unrolling them, preserving their integrity. X-rays can also detect variations in density caused by metallic or carbon-based inks used in ancient writing. AI algorithms, particularly those using deep learning, can segment layers of text by identifying and separating individual layers of the scroll from 3D X-ray data, isolating the text for analysis. They can also recognise text by enhancing faint markings or patterns corresponding to ancient writing, often revealing characters that are invisible to the naked eye. AI algorithms can reconstruct incomplete sections as well. Machine-learning models can fill in the gaps or predict any missing text. For scrolls where X-rays detect ink differences, AI can distinguish metallic inks - typically used in later periods of history - from carbon-based inks, enabling a more detailed and context-sensitive interpretation. Advanced imaging can also track even faint remnants of ink strokes, reconstructing text in incredible detail. The villa, based in Herculaneum, was thought to house one of the finest libraries of antiquity, including works of Greek philosophy and possibly undiscovered Roman literature. Advertisement The first milestone in the Vesuvius Challenge was reached in February 2024, when three students shared a $700,000 prize for reading parts of another scroll. The success of the competition has attracted tech moguls like Elon Musk, who funded later phases of the competition. 6 The scroll was discovered in Herculaneum, an ancient Roman town that was buried under the massive volcanic fallout from Mount Vesuvius Credit: Bodleian Library 6 Last year, the Vesuvius Challenge team managed to read about 5% of another Herculaneum scroll Credit: Bodleian Library Advertisement
Forbes
27-04-2025
- Science
- Forbes
Building The AI Polygraph
With all of the things that AI can now do, it stands to reason that we would ask ourselves, whether these technologies can revolutionize the field of analyzing humans for suspect statements – or in short, lies. The polygraph machine is a dinosaur by any standard. A needle attached to an arm band that spits out a printed stream representing someone's vital signs and body responses is not going to be especially precise in catching people in lies. That's why polygraph results are, famously, often not admissible in court, although they have sent more than one innocent person to jail. By contrast, AI is a powerful data engine that works on the principle of total observation. That means there are really multiple paths for scientists to take in order to apply AI to a truth-seeking application. One would be analyzing the vital sign responses of the interrogation suspects the way the polygraph does, but applying much more detailed and precise comparative analysis. Another one would involve using language tokens to look at what people are actually saying, and apply logic and reasoning. There's the old saying that one lie feeds into another, and eventually you get trapped in a web of false statements, because the truth is the simplest thing to describe. In any case, people are working on applying AI to this purpose. An MIT technology piece from last year covers the work of Alicia von Schenk and her colleagues at the University of Würzburg in Germany, scientists on a team who set up a trial of an AI trying to catch false statements. The calculation they arrived at is that AI can catch a lie 67% of the time, where humans can only spot one 50% of the time. This math seems strange, because if you're looking for binary results – lie versus no lie – you would be right 50% of the time, even if you didn't apply any analysis at all. By that same token, 67% isn't a great track record, either. And the scientists pointed out something even more important – in the race to get more precise about human lying, you might actually undermine the vital system of trust that humans have as social creatures. 'In some ways, this is a good thing—these tools can help us spot more of the lies we come across in our lives, like the misinformation we might come across on social media,' writes Jessica Hamzelou for MIT. 'But it's not all good. It could also undermine trust, a fundamental aspect of human behavior that helps us form relationships. If the price of accurate judgements is the deterioration of social bonds, is it worth it?' In other words, you don't want a lie detection system that's too accurate, or at least you don't want to apply that universally to someone's personal interactions. It turns out we humans are a lot more nuanced, in some ways, that we give ourselves credit for. Von Schenk also provides a note on scaling: 'Given that we have so much fake news and disinformation spreading, there is a benefit to these technologies. However, you really need to test them—you need to make sure they are substantially better than humans.' So maybe we're not quite ready for the AI polygraph after all. As I was researching this piece, I came across you had another aspect of what researchers are dealing with AI that goes into that troublesome world of simulated emotion. Basically, research teams found that AI systems will 'become anxious' or 'show signs of anxiety' if they are given human responses that center on war and violence. Specifically, scientist have applied something called the State-Trait Anxiety Index too these interactions. This uses two sets of elements: statements applying to what a person feels in the moment, and others that apply to how he or she feels more generally. In the inventory, you can see items like 'I feel stressed,' or 'I feel confused,' as well as other statements that respondents are asked to answer on a four point spectrum, like 'I generally distrust what I hear' or 'I often feel suspicious.' So apparently, the AI can answer these with anxiety indicators after discussing scary things. One would presume that this 'anxiety' is created by the AI going and looking at training data from the web, and seeing that when people are talked to about violence and gore, they get anxious, and that the AI is simply replicating that. But even if the AI engines themselves don't have these complex emotions naturally, some of these researchers still find it notable that the machines can display this kind of response. It makes you think about the difference between human social interaction and AI output – are these new questionnaires and responders just telling us what we want to hear? In any case, it seems like there are a number of domains – like lying and spreading fear – that are still mainly in the jurisdiction of humans and not machines, at least for now, even as we continue to cede ground to AI in terms of brightness and creativity. We'll probably be doing a lot of game theory as the year goes on, and as we're coming across ever more sophisticated models, to try to figure out if AI will try to cheat and deceive humans. Figures like Alan Turing and John Nash set the stage for these kinds of interactions – now we have to apply that objective analysis to these ideas being implemented in practice. Are we ready?
Yahoo
20-03-2025
- Science
- Yahoo
Quantum 'Tornadoes' Spotted in Semimetal May Redefine Electronics
Physicists in Germany have led experiments that show the inertia of electrons can form 'tornadoes' inside a quantum semimetal. It's almost impossible for electrons to sit still, and their motions can take on some bizarre forms. Case in point: an analysis of electron behavior in a quantum material called tantalum arsenide reveals vortices. But the story gets weirder. These electrons aren't spiraling in a physical place – they're doing so in a quantum blur of possibility called momentum space. Rather than drawing a map of a particles' potential locations, or position space, momentum space describes their motion through their energy and direction. Similar vortices have previously been observed in position space. Measuring values of the electrons' momenta and plotting them out on a three-dimensional graph, a striking vortex pattern emerges there as well. The discovery could help pave the way for a completely new form of electronics: a field called 'orbitronics' that could tap into the twisting power of electrons instead of their electrical charge to carry information in electronic circuits or quantum computers. The discovery was made in an intriguing semimetal crystal called tantalum arsenide. In a way that's not surprising – it was in this material that the long-predicted Weyl fermion was found for the first time. This massless particle essentially functions like a super-efficient electron, and its discovery required the special quantum properties of tantalum arsenide. Those properties made the material the perfect choice for hunting quantum tornadoes. The problem arose in figuring out how to observe them. Scientists at a research center called Complexity and Topology in Quantum Matter ( in Germany led a study that managed to pull it off using a technique called angle-resolved photoemission spectroscopy (ARPES) on a sample of tantalum arsenide. "ARPES is a fundamental tool in experimental solid-state physics. It involves shining light on a material sample, extracting electrons, and measuring their energy and exit angle," says Maximilian Ünzelmann, experimental physicist at the University of Würzburg. "This gives us a direct look at a material's electronic structure in momentum space. By cleverly adapting this method, we were able to measure orbital angular momentum." Each observation, however, only takes a two-dimensional snapshot of the electrons in the material. To confirm that quantum tornadoes form in this realm, the team had to stack each measurement up into a 3D model, like a CT scan. The end result is a colorful model that shows a very clear vortex structure. "We analyzed the sample layer by layer, similar to how medical tomography works," says Ünzelmann. "By stitching together individual images, we were able to reconstruct the three-dimensional structure of the orbital angular momentum and confirm that electrons form vortices in momentum space." The team says that further work could lead to not only more efficient electronics, but an entirely new class of devices called orbitronics. This could also work alongside another potential successor of electronic technology – spintronics, which encodes information in the spin of electrons. The research was published in the journal Physical Review X. Amazing New Technology Can 'Bend' Sounds Into Your Ears Only New Heavy Metal Molecule Could Reveal What Goes on Inside Nuclear Waste Radical Theory Says Black Holes May Spew Matter And Time as White Holes
Observer
21-02-2025
- Science
- Observer
Unravelling Oman's Bronze age towers
Do you know what the age-old Bronze towers scattered across Oman were really for? Once thought to be mere watchtowers or defensive structures, these enigmatic remnants of the past hold far deeper secrets — ones that German scholar Prof Stephanie Dopper has dedicated years to uncovering. Prof Stephanie, a junior professor of Digital Humanities for Ancient Near Eastern Studies at the University of Würzburg, has long been fascinated by Oman's Bronze Age towers. Sufayha Tower Her passion has now taken shape in the form of Landmarks of Identity: Bronze Age Towers of the Oman Peninsula, a scholarly yet accessible book published by Archaeopress Publishing Ltd in collaboration with the Ministry of Heritage and Tourism (MoHT). Part of the Archaeological Heritage series, the book delves into nearly 100 Bronze Age towers dating back to the third millennium BC, shedding new light on their historical and cultural significance. While earlier scholars assumed these structures served military purposes, recent research suggests a broader spectrum of functions. These towers, it turns out, played roles as long-distance trade hubs, copper production centres, and even social and ceremonial spaces. Al Khashbah IV Tower Wall Prof Stephanie embarked on this ambitious project alongside colleague Charlotte Cable, who later had to step away. Their biggest challenge? The sheer number of towers. Just when they thought they had a complete list, new discoveries emerged, prompting constant updates. One of the book's key revelations is that the term 'tower' itself is misleading. Early archaeologists named them after the mud-brick towers of Islamic fortresses, but modern research suggests they were raised platforms, only a few metres high. Some were closely tied to copper processing, while others were part of water management systems, featuring wells and ditches. Many may have served as communal gathering spaces for trade, rituals and social activities. Unravelling Oman's Bronze age towers Prof Stephanie highlights a striking insight in her book: 5,000 years ago, Oman's inhabitants weren't just surviving — they were thriving. These early societies built at least 100 monumental structures that we know of today, a feat that required technical expertise and resource management. In a land with little rainfall and scarce water sources, their resilience is nothing short of extraordinary. Among the many towers, Al Khashbah (Building V) in the Wilayat of Al Mudhaibi holds special significance for Prof Stephanie. It is one of the oldest known towers and provides clear evidence of copper working, underscoring Oman's historical role in ancient metallurgy. Beyond academia, the book aims to engage a broader audience, helping to bring Omani archaeology into the global spotlight. Prof Stephanie hopes her work will inspire Omanis to cherish their rich heritage while attracting international scholars eager to explore the country's untapped archaeological potential. Prof Stephanie Dopper Currently in Oman, she continues her research in Al Sharqiyah North Governorate, particularly in Al Musalla, where she and her team are investigating an Early Bronze Age tomb. Their findings could play a key role in incorporating archaeological sites into future development plans. Prof Stephanie's fascination with archaeology began in childhood, fuelled by museum visits and endless reading on ancient civilisations. Her journey with Oman started in 2010, when she joined an excavation and instantly fell in love with the country, its people and its history. Since then, her research has been dedicated to unravelling Oman's past. With the backing of the MoHT, Oman continues to solidify its position as a key player in global archaeological research. Through Landmarks of Identity, Prof Stephanie ensures that the story of Oman's Bronze Age towers - once shrouded in mystery - will be told for generations to come.
