Site icon The GDELT Project

Gemini For Museums: Unlocking Centuries Of Forgotten Knowledge Through Archive-Scale AI Translation Of The World's Historical Books

Much of the world's historical knowledge is locked away in the vast rare books and manuscripts holdings of the rare books libraries and archives of the world, either completely undigitized or scanned, but written in long-forgotten languages and scripts largely unreadable to most today. Even as Delrio's Disquisitionum Magicarum's 1,081 pages have been digitized, few today can easily read its mixture of Humanist Ancient Greek and Neo Latin written in a mixture of Greek minuscule with ligatures, Greek majuscule, Latin Antiqua and Latin Italic, while even fewer can read Salomon Henning's Lifflendische Churlendische Chronica's 200 pages in its original Early New High German and Neo Latin in a mixture of Fraktur and Antiqua. Henning's rich first-hand history of Livonia has been digitized, but it remains entirely forgotten and inaccessible due to language and script barriers.

What might it look like to machine translate entire archives en masse from their myriad languages and scripts into modern languages like English, rendering their original texts readable to modern audiences for the first time? Our forthcoming new Gemini Carbon-powered toolkit makes this incredible vision possible for the first time, lowering the costs for advanced machine transcription and translation of entire books spanning multiple languages and scripts in long-forgotten languages to the point where Henning's 200 Early New High German and Latin pages in Fraktur and Antiqua are now available in modern English for a grand total of just $21!

Over the coming weeks we will be demonstrating the potential of this new workflow by mass-translating an entire archive of digitized rare books and other centuries-old works spanning nearly the past 1,000 years, making many of them readable by modern audiences for the first time in history! We see this as an enormous leap in making the world's forgotten and previously lost knowledge accessible for the first time.

Exit mobile version