Artificial intelligence could translate ancient texts

By July 23, 2019 Language, News

Great news, Reading Addicts, your ‘to be read’ pile is set to become enormous as ancient languages could soon be translated.

AI (artificial intelligence) mimics human cognitive functions, one aspect of which is analytical intelligence where the machine can use what it has learned to inform any future decisions. Researchers from MIT and Google Brain have been discovering how to use a machine learning algorithm deep learning to help decipher ancient texts, and read languages that died out centuries ago.

Today there are almost 7,000 languages spoken, read, and understood around the world- this may seem like a lot but this is actually fewer than a quarter of all the languages spoken by humans over the course of our entire history.

While our languages evolve, the structures we use stay basically the same- using similar patterns, rhythms, distribution of characters- meaning if we understood most of the ‘parent’ language, a lost language could be deciphered.

Early Greek language ‘Linear B’ from around 1400BC, was previously deciphered by a human- Michael Ventris – in 1953. Recently, using Ventris’s workings, experts from MIT and Google’s AI lab used machine learning to further decipher the early Greek language Linear B (from 1400 BC) and a cuneiform early Hebrew language shown to be over 3,000 years old.

MIT have been busy translating ancient texts

The experts trained the AI to use 4 key properties related to the context and alignment of the characters to be deciphered – distributional similarity, monotonic character mapping, structural sparsity and significant cognate overlap. By using machines instead of a human brain, the approach can be quick and forceful, with the AI using all its knowledge at once to compare and contrast symbols against each other in order to find similarities. The human mind would take years where the machine can do it in days or hours depending on how large the task is.

With all these new languages being discovered and deciphered, the potential for new reading material is pretty exciting! Thanks to AI and the experts who train them, we could be looking at some insight into ancient civilisations and the stories they would tell each other over the fire.

Leave a Reply