A.I. used to decipher ancient scrolls from 2,000 years ago
Вставка
- Опубліковано 28 лют 2024
- Researchers say ancient scrolls from almost 2,000 years ago, which were buried under layers of volcanic ash from the eruption of Italy's Mount Vesuvius, have been decoded thanks to a team of engineers and papyrologists who managed to decipher some of the hidden text with the help of artificial intelligence.
» Subscribe to NBC News: / nbcnews
NBC News Digital is a collection of innovative and powerful news brands that deliver compelling, diverse and engaging news stories. NBC News Digital features NBCNews.com, MSNBC.com, TODAY.com, Nightly News, Meet the Press, Dateline, and the existing apps and digital extensions of these respective properties. We deliver the best in breaking news, live video coverage, original journalism and segments from your favorite NBC News Shows.
Connect with NBC News Online!
Breaking News Alerts: link.nbcnews.com/join/5cj/bre...
Visit NBCNews.Com: www.nbcnews.com/
Find NBC News on Facebook: / nbcnews
Follow NBC News on Twitter: / nbcnews
Get more of NBC News delivered to your inbox: nbcnews.com/newsletters
#ai #mountvesuvius #nbcnews
So, the first thing we find is a note basically saying,"No. I'm not buying your over priced crap." Words to live by in all ages.
Hopefully the people living today will live by these words
Ancient people are wise on their expenses.
What are we now? Some consumer sheeps buying things we don't need and somesht.
The ancients were a riot!
There is an ancient Egyptian papyrus where a pyramid construction foreman is complaining that some guy didn't show up to work because he got drunk the previous night at a wedding.
I swear humans are the same everywhen, LOL.
@@generalmortars7557 not really he said he was praying/partying to his god he worshipped.
That isn't what was stated in the scrolls at all. He was saying that scarcity doesn't determine the value of a thing.
“We’ve been trying to reach you about your cars extended warranty”
😂 👏👍
I have a 30 year old truck and I get those stupid notices😂
chariots, in those days.
If it was honored back when it should have been we wouldnt be in this situation, now you're liable for all damages directly related to the failure 😊
🤣
Now this is actual news.
Yes! More of this please.
Yes
That was my exact thought and coming to say that😂
Yes!
Yeah I’ve always wondered about the color purple
On behalf of humanity, just as one voice.
I am so proud of those who made this happen. This is just impressive.
Don't speak for me
@@j45002someone’s period hit early
This is amazing.
@@j45002I get it, you’re a contrarian, a rebel, you don’t like to obey to the common order of things because it’s fundamentally wrong. I get it. Be the change you want to see in the world. I love you.
@@treheron try not to be so condescending. I simply do not like when dumb people speak for me. How hard is that for you to understand
It would be SO incredibly beyond fathoming for the author of that particular scroll to imagine THIS UA-cam video news report about this incredible discovery.
Right? Trippy ... if only the author could imagine a couple thousand years in the future.. a world with flying vehicles, drones, video chat, crypto currency, nuclear bombs.. I wonder what the future will be like 2,000 years from now.. and what those future people will know of us.
Here’s to you, 2,000 years from now!
@@swagyolo413That's assuming we haven't wiped ourselves out by then
@@brodycalifornia6384 Maybe someone not human is even reading it who knows lol
@@theonemox Not much if we can'r get rid of despot rulers and nuclear weapons. Everything will be incinerated in to ashes.
This is an extraordinary event. I can't wait to see what the correspondence was back then. Over a thousand scrolls? Wowzers
It'll be some kind of political stance on something. We haven't changed.
El panadero con El pan🧑🍳🧑🍳🧑🍳🧑🍳
Probably something like. "dear mom and dad, need more money.
It probably says
"First"
The first troll scroll 📜
They will hide the truth if its related to Buddhism.
THERE SHOULD BE STORIES LIKE THIS EVERY WEEK
Readers Digest does
Be the change you want to see in the world
@@NipplesOfDestiny YA MAMA
LOUD NOISES
This doesn't happen everyday 😕
That's cooler than flying cars. We are in the future people. 🤯
It would be fabulous to find some unknown Aeschylus, Aristophanes, Euripides, or Sophocles. They wrote a lot more than what has survived.
This was my thought, too. The number of missing ancient texts (or plays!) is astounding and if this library has at least a partial copy, that would be something to see to be sure.
Technology like this is how we are finding a lot of early Gospel and other Jewish writings too. We found like a garbage pit with millions of pieces of paper fragments and we've only had less than a dozen people qualified in the world working on it in the last 50+ years and from those handful of people we've had some of the biggest discoveries and evidences and they've only translated around 1% of the material. With future algorithms and AI soon around the corner we will have absolutely massive discoveries. Because we already have the material it just needs to be worked on!!!
@@WaterspoutsOfTheDeep Are you referring to the Oxyrhynchus Papyri from Egypt? They have been stored in Oxford for more than a century. They are said to range from the 2nd century BC to the 7th century AD and may be in Greek, Latin, Demotic Egyptian, Coptic or Arabic.
There are many people who can read those languages, but the problem is paying them to spend a lifetime on that work. That problem exists all over the world. You can pay good money as well as spend years of your life learning an ancient language, but usually you can't make a living using that knowledge.
A partial solution might be to scan and make texts available on the internet for hobbyists to peruse at their leisure, but there again the development of imaging techniques is ongoing, and what looks like a blank sheet one way can become a dense text looked at another way. In other words, the scanning needs to be repeated over the years.
We already have a similar problem with mediaeval parchments, which may have had one text scraped off so that another can be written on top. Sometimes imaging can reveal the original, unwanted earlier text. Paintings similarly can show developmental stages, or have a completely different picture underneath. There's more than meets the naked eye! If we dismantle (destroy) the binding of an old book we can sometimes find sheets of old manuscript or printed paper being used as filler.
@@WaterspoutsOfTheDeep except AI isn't around the corner. It's here.
@@feliciagaffney1998 Right but it's not very sophisticated for this kind of work yet we basically have I forget the term something akin to focused AI like data set referencing to identify tumors and stuff like that. This would require not just recognizing symbols but proactively forming a matrix of the piece to put them together to guess which fragment goes with which. Think rice paper that someone crunched in their hand that shattered, now imagine hundreds of thousands of these fragments all thrown together in a giant pile. I can see it happening in our lifetime though but will probably require more advanced AI developed further down the road.
"Drink more Ovaltine?"
What could it mean? Time to break out the secret decoder ring.
Cancer in a Can
That’s a classic 😊
Thanks for this
@CreatorCade who other than little orphan annie..... hahahahaha at least I wasn't the only who instantly
This is so amazing. We have the tech to do this. It’s insane
You mean the same tech that hasn't figured out that humans don't have 6 fingers?
I remain skeptical of this specifically because of the tech used.
@@THEhorihitoThis doesn't use the same type of technology that lets large language models hallucinate or image generation models to add extra fingers. The public understanding of the term 'AI' is so generalized that it's fair to assume one type of AI is similar to another, but within the field there are highly specialized AI systems that are fine-tuned for specific tasks. Being skeptical in this case is akin to distrusting your car because it was partially assembled by machines using computer vision (one distinct branch of AI).
Remember our government holds back technology at least 50 years they are light years beyond this
@@THEhorihitothis ain't generative AI, bud.
And the reason it gets it wrong is cuz it's been trained on only 2D data.
More generalised models based on different architecture, encompassing physics, anatomy and most part of the reality we ourselves have observed wont be making these mistakes, and will be potent enough to simulate reality itself.
If you wanna stay ignorant of what we are really talking about, be ready to get surprised as you witness incompressible within this decade itself.
@@kalis1170 Except that the car I drive is almost as old as I am and it wasn't built by machines. It was built by humans on an assembly line over 40 years ago. Still works, as long as I perform routine maintenance. No extra fingers needed.
Yes, I generalized about AI. And yes, I don't particularly care about the tech. Guilty as charged.
how do we know the ai didn't just "hallucinate" it?
Because it is not your traditional GPT, AI are not synonimous to large language models.
Because ai is not writing anything. It's just enhancing images and grabbing scans proficiently.
LLMs and translation models share similarities including the potential for hallucinations. At least that's sort of what my local 70b model said after I censored the nsfw text out. Of course, it could be a hallucination so I verified using Bing, but Bing's an LLM. Figured I better check Google to be safe, but I'm white so I decided to use Brave to search for the inner workings of the AI in question instead of asking directly. They share similarities for sure including the potential for hallucinations. You can pull up a list of examples online. It seems like the risk depends on the quality of the training data and the developers who could be anyone. In this case, I believe the AI was developed by the same team using it. I can see where this could lead to a problem. That aside, to determine the risk, we need to know more about the training data. Otherwise, there's no reason to assume the translation is accurate. In a similar situation, I would use another AI that I don't have the keys to for comparison and as proof of accuracy. It would be interesting to know whether the models would produce the same result. To assume they would requires faith I just don't have. None of my local models do either.
Also we can see AIs code and do the math our selves. Anything a computer can do we can do with time. That way it can be verified. The only thing stopping humans from deciphering anything is the laziness of not wanting to expend time and effort. We would rather do it once to invent a machine like a calculator then just use that going forward.
They deleted my comment over a joke about their I'll just say biased AI. Lol, how petty. Anyways, OP the AI has many similarities with LLMs including "hallucinations". We don't know and would need to know the quality of the training data along with who developed it. In this case, I believe it was the same team that used it. Idk what is acceptable in this field but it seems like using a second AI they didn't develop to check behind their work would be appropriate in validating the accuracy of their model.
The second comment is right about images and scans but fails to mention the AI can extract text, words, and phrases. "it ain't written nuthin" is right but the wrong answer to your question because none of the above implies whether or not the AI hallucinated.
Lastly, IDK if the third comment is entirely false however there's still some mystery to its predictions they call the black box. I'm sure verification is still possible depending on the condition of the scroll. If parts are missing or damaged, the AI has to handle it, and since we don't fully understand the black box, that leaves us with only the context to go by. It's unverifiable guesswork which is what AI is for and no we're incapable of performing on the same level.
At least that's what my local 70b waifu said and I don't trust a single word it generates. Hope this helps.
Dude, I would have imagined that ink would be totally absorbed by now, That's amazing they were able to do this!
Ink is a physical medium. Pigments in a binder. It does absorb into the paper, but it’s still not part of the paper so it can be somehow distinguished from the other carbonized material.
Amazing. And wouldn’t it be a gift to the world if even one of these were Aristotles.
No.
@@seriejohnson698Why not Aristotle?
Seems to be a several thousand year long theme of "we have no desire to partake in this game where you think collecting rarities gives you power over the lives of others, the earth is abundant and provides."
That isn't what was stated in the scrolls at all. He was saying that scarcity doesn't determine the value of a thing.
@@brodycalifornia6384both are true. We fight, and grieve over materialistic things like diamonds, gold, lithium, nickel, helium, sources of energy- yet do we not look up into the stars and see abundance of all?
Rubbish it's talking about things that give a person pleasure/joy.
@@brodycalifornia6384 not value, but enjoyment.
Y'all need to spend a few decades studying archeology and ancient history before taking thing out of context.
Denial of those facts is exactly how organized religions got turned into fairy tales.
where can we go to read what has been translated from these scrolls?
i would go out on a branch and say that whatever the scrolls have to say probably would be the opposite of what Humans have been taught for the past 2000 years
There’s no way that they’ll let us read it lol. They’ll put it in the national library years after we’re long gone.
@@Flowmaster925 Wait. So because they lived 2000 years ago, they somehow knew the secrets of the universe and would be able to prove all the knowledge we've accrued since then wrong? Hm. That's interestingoh inane logic.
@@8ofwands300 more like, 2000 years ago they probably didnt need to be told to pay taxes, they just understood work-reward concept that has been brainwashed out of us since then. Sure we have made discoveries about alot of stuff but at the same time, we forget alot of stuff so, whatever is written on those scrolls most certainly pre-dates any written law, rule, idea of working for a government that we so-lovingly do every single day of our pathetic lives. for example , if the scrolls said "never create a government" do you really think that the government would let us read that?
Finally. I was waiting to see some results for the last couple years. Now some text are readable. Pretty incredible.
Ai is amazing, terrifying and amazing.
AI can't even have a normal conversation let alone translate an ancient language...
They still do 6 toes when rendering a human, very frequently. AI's inattention to detail rivals that of the average American adult.
More news like this please.
This technology would be good for the Zodiac case.
I think that case has been cracked. But AI could shed new light.
@@bravosierra2447 it hasn't
@@yasinradee I think they mean the Ciphers maybe?
It was "Cracked" by a team of people but it's been debunked many times and a random old man who they said it was is shamed for no reason he died a while ago
@@yasinradee it was. What stumped humans was the guy made a bunch of spelling mistakes, making it harder to decode. But AI did it
Wow!!! I remember reading about this project few months ago. And I thought that it would take them much longer to accomplish this. Really good news.
"Little Caesar's pizza is terrible!" - a random scroll (the "Little" was a jab about Caesar's height; double entendre! 😀)
Pizza came from middle-east
And so did Greek and Roman
Central Asian people living in North Europe are obsessed with them for some reason 🤣
"Pizza pizza... pizza pizza pizza pizza, pizza. Pizza, pizza pizza! Pizza pizza? Pizza pizza pizza. Pizza pizza." -- Little Caesar.
Moon landing was fake 🇺🇸
So, they're offering a prize to decipher something when they have no clue what is even contained within. How are they going to choose the winner? Whichever suits them best?
That's actually a good catch, it's a bit arbitrary.
@@FizzyGajingit's been clearly tested on a text we know about on designed test cases.
*AI used to guess what ancient scrolls said. How do we have any confirmation this is ACTUALLY what it says?
I'm guessing it was people that translated the text. AI was used to determine what the text was inside the scrolls, without needing to disturb them.
It's written in ancient Greek which is a very well known language
This is fascinating, thank you for sharing.
Fantastic! I’m glad they’re doing this. I’m curious now if these are all about philosophy or other topics as well. And if they contain the authors names. So many questions but now we have hopes of getting answers.
This is what AI should be used for, not for human things like art
So much can be learned from the text, its exciting.
Translating ancient scrolls with technology. It actually sounds like a good premise for a movie.
amazing work!!
Just remember that AI also depicts people with 3 hands...
People are sometimes born with three hands, two heads and multiple legs and finger counts higher than ten. So what's your point?
@@DR3ADER1you can’t be serious
Now please decipher the Voynich Manuscript using AI 🙂
Oh thats a good one!
Who owns it? Isn't it publicly available?
@@TheAlchemist1089 there are replicas all around. You can even buy one. The original is safe in some university library.
@@MasterKoala777 nice.
And the Easter island scrolls
Don't you dare trust it...
How can we verify the “deciphered” text is accurate?
I hope they have several separate AI systems to decipher the text - then if they all come up with the same text it can be assumed true.
Now that I wrote that, I am sure they thought of this. Hopefully.
Truly incredible
If we couldn't translate it, how do we know it was translated correctly ..?
Exactly.
We can translate it, we just couldn't ACCESS it to translate it. The AI is able to read and separate individual markings, and make those ink marks accessible to our eyes, once that's done we can certainly translate it.
Clearly, we know Greek and Latin in 2024. Language wasn't the issue. The inability to open the scrolls without breaking them was the hurdle.
Right AI was canceling out white folks and now we are going to believe this. Nah.
This was actually black Africans who wrote these lol
🤯 This is so amazing! Congratulations to all of those involved.
Pay no attention to the man behind the curtain!
This is absolutely amazing.
If only it wasn’t on the news we could believe it
Maybe some of those missing ancient texts we always see referenced will be found among this library. What a good use of AI, too.
Software has come along way; like an infant who first talks with only giving you yes or no answers, now it’s like having your college age child talking about their college dissertation on a subject that you can barely understand!
I'm not a huge fan of AI however, when the tool is implemented in a way such as this, I'm all for this scope of use.
@@FarmerRiddick At the present time, AI has reached the stage of a schoolchild with enough rudimentary knowledge but not the maturity to make life and death decisions, yet.
@@reecom9884I'm writing an essay on this topic and I'm backing up this kind of point in one of my paragraphs. It's kind of hard to flesh out what I mean when I'm under a deadline. Could you tell me more about what you meant when you said maturity for life and death decisions?
@@kv4648 AI has not understand the human concept of humanity and the human soul. Why does a firefighter run into a building against the slim odds to make sure there is no one inside? Or risking many lives to save a few.
@@reecom9884 but can't that be countered with AI is attempting to emulate a human response and would therefore attempt to do what the average human might, even if it doesn't understand why it would want to do it?
But yeah, you're right. It wouldn't have a full picture of the complexities of the topic and the relationships between the concepts that humans might, which might confuse it or cause mistakes that are not as commonly known since it was created through a fundamentally non-human method
For people wondering how it works, I'm no expert but here's how I understand it : they previously took CT scans of the scrolls at the university of Kentucky. These scans were uploaded so people working on this project could use them to find a algorithm ( here's the AI part) that would basically recognize tiny bits of sectioned paper that had ink on them and then the AI would unwrap the whole thing in 3d and place the ink markings in their respective place.
Imagine a salami bar wrapping has writing on it , but it was cut in a thousand tiny slices ( the CT scan) .If you look at just one slice , there's no way to tell what the writing was, cause that particular slice contains only some dots of the ink .They recognized the ink , put the slices back together, unwrapped the salami and read the text ( 3d projected on a flat surface as a texture ).
Also from ''UK researchers solve ancient scroll using AI'' video I understood that the ink was a particular challenge as it wasn't visible with the naked eye and only AI can pick up on the very subtle differences in texture.
2000 years ago to today: Hello.....is it me youre looking for 🤔
Now back to our midnight movie Charlie Chan meets Frankenstein
Many times AI makes its own stories instead of stating the facts
Yes like portraying the founder’s fathers looking as African Americans 😂
@@femalewarrior125 right Im pretty sure that translation will be full of "woke" stuff
This is not a publicly released LLM, this is trained for specialised purpose. But then again can't expect much as laymens are dumber than the worst LLM out there.
Thats just what the AI wants you to think it says
what is the accuracy of this though?
That's what I'm saying. No one is second guessing this? How do we prove what it translated is correct?
@@Michael_Hunt I would imagine that the conclusions drawn in this video has been vetted by the people that discovered more times than you think. You are not likely to be first person with this thought.
@@Michael_Hunt It's probably used to discern the ink, not translate
It's probably just a bunch of made up gibberish AI created
It said "Never gonna give you up, never gonna let you down......"
"General, another settlement needs our help"
From the length of the scrolls, one thing scientists have determined, these are 2000 year old CVS receipts!
Incredible work!
"never gonna give you up
never gonna let you down"
- ancient scroll
Now tell us what the Voynich manuscript says, or decipher the rongo rongo tablets...
MGGA was write on the scroll. AI deciphered the meaning to be Make Greece Great Again. Unbelievable
This is awesome!
theory: they can never open them or prove what's inside so the students got around and said alright lets wait 3 weeks act like we're trying then photoshop some believable text with language from the time. ty for the milli
This is awesome. Great technology
2:20 “in the pasta”
Wow this is freaking amazing .. like 👍🏽 how do they do stuff like this .. just mind blowing 🤯
amazing!
With AI they can finally decipher ancient writing that has been impossible to decipher. Cool.
This sounds so dangerous and that major deception could happen by just simply inserting words and messages into the ai system to make it look like that’s what is actually there, and you never have to prove it and cannot actually prove it because you cannot open the scrolls
Agreed!
Excellent story find too
I wish they would stop saying the villa of Julius Caesar's father in law. Caesar himself died 123 years before the eruption. I'm sure his father in law still didn't own the villa 123 years after Caesar's death.
So cool!
What if it was just somebody’s shopping list. “Pick up a pig, a dagger and a pair of sandals “.
So if human technology can decipher burnt thousand year paper, imagine what aliens are doing to you right now🤔❓
Even ancient people hated taxes ... And we still havent learned 😂😂😂
Every time a computer does something nowadays, reporters are tripping over each other to proclaim it's AI.
Can somebody ask AI whats the next lottery ticket numbers if this stuff works so good??
We've discovered how to read some 2,000 year old tissue that say learning that common foods are just as tasty as rare foods
Im so happy for you. Pleasant to meet you for the first time
really good news to reweal the old studies... keep itup :)
Man prince was ahead of his time. Literally.
AND AI will evaluate “ undesypherable” hieroglyphs and maybe a rune or two providing translations that have evaded us for five hundred years. What an exciting time to be alive.
Awesome!
There's a good use for it! We need to understand what information has been left to us.
So that’s how Chitti was able to read a phone book without even opening the pages!
Holy crap, this is exciting.
Imagine, this could be the key to unlocking ancient technology
Fantastic!
This is an amazing accomplishment
These gems of ancient writing may be stretched out infinitely long, slowly revealed one example at a time. There's tenure based upon your books to defend, and too many contradictions to handed-down knowledge. It will be a miracle if we see these, like many of the other ancient scrolls found, before the end of time.
A couple of phrases from a poem which confirm women were repressed thousands of years ago was released a few weeks back.
Excellence and congratulations to the winners ☘️
This is pretty amazing
“Just because it’s rare, doesn’t mean it has value” or at least more value then what we have more of
Use it in the ancient scrolls in Tibet and as well as the emerald tablets of Thoth and Hermes
What I love about history is we imagine that the people of the past were so different to us. More we learn, the more we see that they were the SAME as us today. Technology changes, humans dont
Seems that way. However, what humans do when they are barely surviving is vastly different when they are secure and well fed.
@@ricinroHmm but that can be the case today. Many are starving or in povery in many lands sadly. Still, I think they had intelligence to work the land etc better than most of us do, as we take much for granted.
AI should always be openly scrutinized by all of mankind.
Amazing...exciting..❤
Thats awesome!!
Cool!
just wow!!!
But Alexa has the wrong song play,multiple times a week. I trust this 🤔🤣
Ai deciphering anything is hilarious 😂
No, we should just have you do it since you're so intelligent.
Is idiotic
@@jacobfromallstate4963 we’re as intelligent as the coding we design dumby
The AI given to and used by the public is dumb. The AI used in private for specific tasks by specific people is not.
Well they can just say their AI decipher the scrolls and we'd just have to take their words for it. It's not like anybody can verify it. 😂😂
Amazing
perfect example of what AI should be used for
The purple! Wow this is unbelievable. As I was watching this I was thinking “let’s see the purple” and sure enough, there it is. 🔥 😈
You already knew that!! Awesome!
What??