Scroll Burned in 79 AD Volcanic Eruption Finally Deciphered Using AI (smithsonianmag.com) 50
When Mt. Vesuvius erupted in 79 A.D., it buried hundreds of papyrus
scrolls. They were rediscovered in the mid-1700s, remembers Smithsonian magazine, "the only
surviving collection of its kind from the Greco-Roman
world..."
"But when scholars tried to unroll them, the carbonized manuscripts crumbled to dust." Every generation that followed faced the same dilemma: They could wait for technology to advance, abandoning hope of reading the ancient texts in their own lifetime. Or they could try to open the scrolls themselves — and risk destroying them.
In recent years, researchers have settled on a third option. Using advanced imaging and artificial intelligence, they're deciphering the scrolls without needing to unroll them at all.
The Vesuvius Challenge has accelerated the process by turning it into a public competition, complete with cash prizes. In 2023, a student won $40,000 for deciphering a single word — "purple" — from an unopened scroll. Later, contestants would identify 2,000 Greek characters from one scroll ($700,000) and the title of another ($60,000). Now, for the very first time, researchers have recovered all surviving text from a single scroll. The nearly five-foot-long segment includes roughly 20 columns of ancient Greek philosophy, accessible for the first time in nearly 2,000 years.
"The tech actually does look like magic, but it's not," Brent Seales, a computer scientist at the University of Kentucky, said at a press conference. (The article points out that Seales partnered with two Silicon Valley investors in 2023 to launch the Vesuvius Challenge, and is now hailing "the restoration of lost voices from the ancient world." Seales has been working on virtually unwrapping the scrolls since the early 2000s. The process involved imaging the bundles of papyrus using technology similar to CT scanners, isolating thin layers and then stitching them together.... "We've developed a systematic and a repeatable approach," Seales told the audience. "Now it's only a matter of time until we read all of the scrolls."
"But when scholars tried to unroll them, the carbonized manuscripts crumbled to dust." Every generation that followed faced the same dilemma: They could wait for technology to advance, abandoning hope of reading the ancient texts in their own lifetime. Or they could try to open the scrolls themselves — and risk destroying them.
In recent years, researchers have settled on a third option. Using advanced imaging and artificial intelligence, they're deciphering the scrolls without needing to unroll them at all.
The Vesuvius Challenge has accelerated the process by turning it into a public competition, complete with cash prizes. In 2023, a student won $40,000 for deciphering a single word — "purple" — from an unopened scroll. Later, contestants would identify 2,000 Greek characters from one scroll ($700,000) and the title of another ($60,000). Now, for the very first time, researchers have recovered all surviving text from a single scroll. The nearly five-foot-long segment includes roughly 20 columns of ancient Greek philosophy, accessible for the first time in nearly 2,000 years.
"The tech actually does look like magic, but it's not," Brent Seales, a computer scientist at the University of Kentucky, said at a press conference. (The article points out that Seales partnered with two Silicon Valley investors in 2023 to launch the Vesuvius Challenge, and is now hailing "the restoration of lost voices from the ancient world." Seales has been working on virtually unwrapping the scrolls since the early 2000s. The process involved imaging the bundles of papyrus using technology similar to CT scanners, isolating thin layers and then stitching them together.... "We've developed a systematic and a repeatable approach," Seales told the audience. "Now it's only a matter of time until we read all of the scrolls."
Let me guess (Score:2)
Re: Let me guess (Score:2)
Re: Let me guess (Score:2)
Re: (Score:2)
Nope:
Be sure to drink your Ovaltine.
Re: Let me guess (Score:2)
Re: (Score:2)
Rickroll FTW.
"Never gonna give you up..."
Re:Let me guess (Score:4)
Elevators had been invented and were already used by the Romans in various places.
However, it would be unlikely that these warnings would have existed in their apartment buildings (insulae). The risk of fire was extremely high compared to today (open flames used everywhere), and the well off residents certainly wouldn't have lived in penthouses encumbered by stairs. The most desirable apartments were actually located at ground floor, which was easier to run out of. The top floors were generally occupied by the poor who didn't matter.
Re: (Score:2)
Re: (Score:2)
It's common knowledge. Maybe not for you though? I'd suggest taking a class in ancient history to close those educational gaps.
There are at least three classes of supporting evidence: literature (eg Juvenal, Cicero) ancient ruins (eg Pompei, Rome) and modern historical research. Here's a popular science book [amazon.com] for you.
How will our encrypted USB devices be read? (Score:2)
Re: (Score:1)
Re: How will our encrypted USB devices be read? (Score:2)
Re: (Score:2)
At best you might get a mysterious single damaged disk drive, not unlike the Antikythera mechanism [wikipedia.org]. But unlike Antikythera, deciphering the physical mechanism is not enough, archaeologists will have to figure out the magnetic encoding, perform error correction on the deg
Re: (Score:2)
That reminds me of the process at UC Berkeley, using lasers to read and preserve wax-cylinder phonograph recordings. Modern analysis tools can scan, image, and decode earlier technologies.
Project Irene [berkeley.edu]
The FAA is also removing accident investigation reports that show spectrogram images of the Cockpit Voice Recordings. Someone has analyzed the images and recreated the voice recordings, accurately enough that survivor's families could recognize the voices. Newer technologies and computational resources will
AI or Machine Learning? (Score:3, Insightful)
Seales has lately been enhancing his technique, by using artificial intelligence to train his software to recognize subtle differences in texture between papyrus and ink. He plans to combine such machine learning and X-ray fluorescence to produce the clearest possible text. In the future, “it’ll all be automated,” he predicts. “Put it in the scanner and it will all just unfurl.”
What's the difference? (Score:3)
Seales is using conventional machine learning techniques that the article's author has conflated into the buzzword of the day, artificial intelligence.
Machine Learning *is* an AI technique. Artificial Intelligence has always been an umbrella imprecise word to define a large class of very different algorithms, with the only trait in common that the programmer doesn't know a precise sequence of steps to find the solution but rather the data-driven program explores the inputs to try and find useful results.
Say
Re: (Score:2)
It's not wrong, but it isn't right either. AI has gone through several boom bust cycles since the first neural network was invented in the late 1950s, and no doubt will go through several more in the next 80 years. Currently we call everything AI because it allows money to flow into projects by linking them with the stratospheric stock valuations of AI tech companies. Success loves company and all that. Conversely, when the next bust cycle hits, AI will become a dirty word and projects will go out of their
Re: (Score:2)
Machine Learning is a subset of AI. Nope, it's not the other way round, even when some people assume that.
Re: (Score:2)
You are correct. However, I think crunchy_one has a point. TFA says:
Seales has lately been enhancing his technique, by using artificial intelligence to train his software to recognize subtle differences in texture between papyrus and ink.
That sounds like Seales is using data to train a model. And that in fact is machine learning. So, the writer of TFA apparently did "conflate" the two terms.
Re: (Score:3)
Yeah, but ML is a (true) subset of AI, so that's a correct (but not the most specific) statement.
People in Slashdot comments are picky, but in science papers the term AI is very often used as broad umbrella and the term ML seems to be currently basically nonexistent. I guess part of the problem (if you see a problem) is that machine learning is clunky, ML no well-established acronym and AI (rarely spelled out) used since decades.
I am also not sure why you think "train a model" would change things. You almos
Deciphering documents you can't view first-hand? (Score:3)
Seems like perfect circumstances for using hallucination-prone techniques!
#whatcouldpossiblygowrong
Re: (Score:2)
Re: (Score:3)
omnis tuus basus est belongus ad nos
Re: (Score:3)
From the past (Score:2)
Fascinating, when could we read the scroll?
There are also lots of artefacts from the last century lost. Video games from the 1980s.
One could recreate music by optical 3D scanning shellac records and combining different prints to eliminate the noise. The quality gets better. Just image a new scan of Metropolis negative with current technology Or Robert Johnson even better than in the Centennial edition..
Re: (Score:2)
>Video games from the 1980s.
Probably not all of them, but it is surprising how many ROM images exist out there and how available emulators are to run them on modern computers, if you look hard enough.
I've seen every game I ever played on my C=64, and I've seen a lot more for every console I've ever heard of except ColicoVision. I assume those exist as well, somewhere.
Re: (Score:2)
Reconstruction of Mechanically Recorded Sound by Image Processing (2003) https://www-cdf.lbl.gov/~av/JA... [lbl.gov]
Damnit (Score:2)
Now, for the very first time, researchers have recovered all surviving text [scrollprize.org] from a single scroll.
I was betting on one of these [fandom.com] ... :-)
I deciphered one too (Score:5, Funny)
LISTA EMENDORUM
(Marcus needs to pick up before Livia kills him)
Garum (the good Pompeian stuff, NOT that cheap Lusitanian rubbish)
More garum (you can never have enough garum)
Olive oil, 1 amphora (check for watering down, that crook at the forum did it last time)
Bread, 3 loaves (the ones without the sawdust)
Dormice, 12, fattened (Decimus is coming for dinner, the pretentious bastard)
Flamingo tongues (see above re: Decimus)
Posca for the slaves (vinegar will do, stretch it)
Wine, Falernian if we can afford it, Campanian if Livia isn't looking
Lead acetate wine sweetener (everyone says it's fine)
Silphium (if anyone still has any, which they don't, because you lot ate it all)
Fish, whatever looks least suspicious
Lark tongues, 200 (Decimus again, honestly)
Snails, 1 bucket, milk-fattened
Cumin, because we put cumin in everything
Pepper (remortgage the villa first)
One cabbage (Cato says it cures everything, Cato is a bore but it can't hurt)
Urine, 1 amphora (the fuller needs it for the togas, don't ask)
Sponge on a stick (we're running low and the public ones are disgusting)
If there's change left, new toga. The old one has garum on it. Again.
Output (Score:3)
"Warning from the Star Visitors: don't entrust society with mechanized thinking devices."
The bad news... (Score:3)
...is that after reading the scrolls we can confirm what we already knew, that these people were stunningly ignorant by today's standards. They don't have anything to say to us.
Maybe future scrolls will be better... (Score:3)
Turns out this one was a CVS receipt.
Re: (Score:2)
deciphered or interpolated/hallucinated? (Score:2)
I know from tesseract and other "modern" neural-network OCR software I had the displeasure to work with that they basically guess stuff based on the data they were trained on, rather than from the data encoded in the picture they're supposed to "read". When meeting unique words or dialectal forms not present in their training data, they confidently replace it with some plausible shit, rather than the gibberish the old-generation OCR were producing.
If those guys were really able to virtually unscroll those b
It's not the third option (Score:2)
OK, it's only a niggle, but:
"They could wait for technology to advance, abandoning hope of reading the ancient texts in their own lifetime. Or they could try to open the scrolls themselves — and risk destroying them.
In recent years, researchers have settled on a third option. Using advanced imaging and artificial intelligence, they're deciphering the scrolls without needing to unroll them at all."
Nope, "advanced imaging and artificial intelligence" is the advanced technology for which they were waitin
Has the copyright expired yet? (Score:2)
What a way to honor Chrysippus (Score:1)
AI advertisement again (Score:2)
Validation (Score:2)
AI spits out text it says came from the scrolls. How is anybody going to verify that it got it right?