Cuneiform: How AI is revealing the secrets and techniques throughout the world’s oldest texts

BEHIND a locked door within the British Museum, London, there’s a lovely library with excessive, arched ceilings. Inside this secret room, Irving Finkel opens a drawer and pulls out a clay pill. Cracked and burnt, it’s imprinted with the tiny characters of the world’s oldest written language. It’s a checklist of omens. One other drawer reveals one other pill. “This is a prayer to the god Marduk,” says Finkel, who’s assistant keeper of historical Mesopotamian script, languages and cultures on the museum, and one in all solely a handful of individuals on the earth who can learn this long-dead script, often known as cuneiform, fluently.

Behind us, a photographer is meticulously capturing photos of this writing, with lights positioned to spotlight the indented etchings. This work is a part of a revolution, one that’s utilizing right this moment’s computing energy to carry this 5000-year-old document again to life and unlock new secrets and techniques of the world’s first civilisation.

Though this method of writing was deciphered 165 years in the past (See “Studying the indicators“), the vast majority of texts that use it have by no means been translated into fashionable languages – a fiendishly difficult process that depends on specialists reminiscent of Finkel. Now, because of developments in synthetic intelligence, computer systems are being educated to learn and translate cuneiform, to place fragmented tablets again collectively to recreate historical libraries and even predict bits of lacking textual content. These instruments are enabling the earliest works of literature to be learn in full for the primary time since antiquity, giving insights into tales that later appeared within the Bible and shedding mild on civilisations on the daybreak of historical past.

The story of …