It's not recreating history, it's simply clarifying the images so the original text will be legible. Sheesh. Less paranoia, okay? Or do you think CT scans and MRI's recreate your body?
Im so glad you provided the example of CTs and MRIs, otherwise my remedial brain wouldnt have been able to understand what you meant. And yes id be opposed to that if they were used to explain the evolution of the human body or any historical context whatsoever.
How about photographs? They can have their colors manipulated, be faked, used to image things the human eye can't even see, like UV light. I think it’s not the technology, but the fakery and our worry about being manipulated by an image, a text, or a machine. I think you're worrying about skullduggery in the wrong place, though. It's the humans you have to watch out for ... please don't throw out the useful parts of the baby technology with the bathwater that may still seems a bit murky.
Of course scanning technology -- which involves things like e.g., noise reduction, false coloring -- is used to scan fossils as well, and thus derive evolutionary insights. Your objections make no sense.
Do we reject germ theory because human eyes can't see things that small and we have to use a microscope? Do we stop using bulldozers to dig because they are stronger than we are? Lasers because they produce an artificially phased light? Dyes making colors that don't occur in nature? We use so many inventions to do things we really want or need because without them we could not. CT scans to find cancers, or proton therapy to treat them. Satellites for cell phones, and computer-guided rockets to launch them as no unassisted human could do. In this case, we really, really want to find out what these incredibly damaged Greek Roman papyri say that would otherwise be lost to history. This new tech is not re-writing history -- it's just letting us finally discover it. More Dead Sea Scrolls next?
AI is transforming the translation of ancient Babylonian tablets by enabling faster, more accurate interpretations of cuneiform texts. Machine learning models, such as those developed by researchers at Tel Aviv University and Cornell, can now translate Akkadian with high accuracy, particularly for administrative and royal texts.
This story is human collaboration at its best, a multi-year, multi-organization, amateurs and experts converging on a shared goal. And it beautifully illustrates what human-AI partnership actually looks like when it works.
The ground truth problem is telling: without knowing what we're looking for, how do we know what to attend to? William James observed that experience is what we agree to attend to. That's exactly what's happening here, researchers are teaching machines which ink traces matter, which surface distortions are meaningful, what deserves attention.
Perhaps the real breakthrough will come not from more data, but from systems that can develop their own theories of what's worth noticing. But I hope we resist the urge to fully automate humans out of this process. Every character recovered from these scrolls is worth millions of lines of today's daily linguistic output. Some things deserve the slowness of human attention.
Truly an inspirational story, thank you for sharing it.
Not a fan of having AI “recreate history” artificially … a slippery slope
It's not recreating history, it's simply clarifying the images so the original text will be legible. Sheesh. Less paranoia, okay? Or do you think CT scans and MRI's recreate your body?
Im so glad you provided the example of CTs and MRIs, otherwise my remedial brain wouldnt have been able to understand what you meant. And yes id be opposed to that if they were used to explain the evolution of the human body or any historical context whatsoever.
How about photographs? They can have their colors manipulated, be faked, used to image things the human eye can't even see, like UV light. I think it’s not the technology, but the fakery and our worry about being manipulated by an image, a text, or a machine. I think you're worrying about skullduggery in the wrong place, though. It's the humans you have to watch out for ... please don't throw out the useful parts of the baby technology with the bathwater that may still seems a bit murky.
Of course scanning technology -- which involves things like e.g., noise reduction, false coloring -- is used to scan fossils as well, and thus derive evolutionary insights. Your objections make no sense.
Yes, it should made up by humans. 😉
Do we reject germ theory because human eyes can't see things that small and we have to use a microscope? Do we stop using bulldozers to dig because they are stronger than we are? Lasers because they produce an artificially phased light? Dyes making colors that don't occur in nature? We use so many inventions to do things we really want or need because without them we could not. CT scans to find cancers, or proton therapy to treat them. Satellites for cell phones, and computer-guided rockets to launch them as no unassisted human could do. In this case, we really, really want to find out what these incredibly damaged Greek Roman papyri say that would otherwise be lost to history. This new tech is not re-writing history -- it's just letting us finally discover it. More Dead Sea Scrolls next?
The sheer stupidity lol. If you hate this so much, maybe you shouldn't read a substack titled, "Understanding AI"
Amazing anyway.
And what about that? 👇
AI is transforming the translation of ancient Babylonian tablets by enabling faster, more accurate interpretations of cuneiform texts. Machine learning models, such as those developed by researchers at Tel Aviv University and Cornell, can now translate Akkadian with high accuracy, particularly for administrative and royal texts.
I wish Farritor had stuck with this very impressive work.
This story is human collaboration at its best, a multi-year, multi-organization, amateurs and experts converging on a shared goal. And it beautifully illustrates what human-AI partnership actually looks like when it works.
The ground truth problem is telling: without knowing what we're looking for, how do we know what to attend to? William James observed that experience is what we agree to attend to. That's exactly what's happening here, researchers are teaching machines which ink traces matter, which surface distortions are meaningful, what deserves attention.
Perhaps the real breakthrough will come not from more data, but from systems that can develop their own theories of what's worth noticing. But I hope we resist the urge to fully automate humans out of this process. Every character recovered from these scrolls is worth millions of lines of today's daily linguistic output. Some things deserve the slowness of human attention.
Truly an inspirational story, thank you for sharing it.
Thank you for informing us about an application of AI that is not focused on surveilling and/or deceiving us for profit.
PBS aired a documentary film about this: https://www.pbs.org/wnet/secrets/preview-the-herculaneum-scrolls-8kazks/8065/
Thanks for the share