The AI historian: A new tool to decipher ancient texts

The origins of ancient inscriptions are often shrouded in mystery. Writing carved into stone millennia ago can be hard to read and is often missing entire sections of the text. Now a neural network, trained on thousands of existing inscriptions, could help historians figure out when and where a piece of writing comes from - as well as fill in missing words and characters.

[inchlegendaka.icebrrg]

My handwriting is the final frontier when it comes to desciphering ancient texts, I am willing to submit a scientific sample for research purposes

[RicardoBatistaPassos]

Thank you Nature for sharing knowledge through such well made edited videos!

[Ristro]

It's really cool that it gives you multiple options and not just one that's the 'most probable'.

[silviaholm]

this is an excellent tool for everyone who works with inscriptions, I hope it will be extended to other languages and scripts as well!

[Nicholas.Tsagkos]

Kudos to both of you, excellent work and huge offer to the cultural legacy of Greece.

[DonatellaNellieFatou]

Great work 👍
Best wishes to team Ithaca 🌹

[KenDebono]

Would be nice to feed ITHICA the Rongorongo tablet's of Easter Island as well as the Voynich Manuscripts. It would be cool to have a variant of ITHICA to decipher crop circles.

[joshuapatterson]

This guy is awesome!!! Yannis needs a show

[oldsailor]

Fascinating project with lots of potential!
Can ethaca be fed by image recognition? I feel this will open up great horizons, and not only for greek inscriptions.

[TheGreatRoja]

I'd love to see this applied to The Dead Sea Scrolls

[rilosvideos]

Thats a great usecase for AI! There are many ancient languages/writing systems still completely undecyphered, like Linear-A and many others. Hope they will be readable some day.

[peter.biedermann]

Absolutely marvellous! Thank you for bringing together the fields of sciences with that of the humanities.

[danielbruceagra]

One thing that might be interesting is using the IA to identify patterns in the Linear A to solve the minoan language, like, words that come first, grammatic, and other things like that, could be very interresting

[koumospecial]

This is truly amazing 😮!!!! Congrats 👏👏👏👏👏👏👏👍👍👍👍👍👍

[ihtesham_emon]

All of Deepminds projects are mind-blowing, this not an exception 🔥

[DeepFrydTurd]

All you need is a data base full of all the ancient text identify every character then just ask the computer what it says. It's not hard. You see it in captchas all the time. The hardware is the expensive part lol an also the human individuals that will translate every character and the Python and C++ and Java programmer that will make the interface.

Cloud GPU Azure or Google = ML Hardware
Hieroglyphs expert translator = PhD
UX design expert / programmer = converting ML cloud output to App or Software
Now how to aquire the ancient text and scripts 🧐

[DeepFrydTurd]

How the hell do I get the scripts and hieroglyphs. The rest is easy just need the ancient scroll's

[kickassvideos]

am i wrong or could you not nearly archive the same with simple NLP algorhythms like bag-of-words, word2vec, n-grams etc?!?! with those algorhytms you tell with high proberbility what words are before and after the word "delphi", or what is the missing word, you could also say how the words are related to each other, like king -> male etc.

[jperez]

the clay tablets from all museums should be photographed and 3d scanned and uploaded for anyone to use and decipher

[billbuyers]

I wonder if anything more has come of this. Or is it really being used at all?