Cracking the code: can AI help us decipher ancient languages?

According to The Conversation, ancient languages like Linear A and Etruscan have defied linguists for over a century because they lack linguistic relatives or bilingual texts to guide decryption. In June, a self-taught AI engineer made a high-profile attempt to crack Linear A by testing a hypothesis about Semitic roots, reportedly identifying forty sign values and compiling a four-hundred-eight-word lexicon with machine learning assistance. But here's the tension: AI is powerful at rapidly cross-checking patterns across entire archives, work that would take months by hand, yet it cannot manufacture meaning from statistical analysis alone. The technology requires an anchor—a known language family or bilingual reference—to tell which patterns are meaningful and which are coincidence. With only seventy-five hundred surviving characters, Linear A's tiny corpus makes almost any hypothesis find scattered supporting matches. Until a genuine comparative anchor emerges, AI remains what it's always been: a supercharged research assistant to fundamentally human linguistic puzzles.

Listen to this story

Hear this and more stories in a personalized audio briefing.

Open The Chonkerton