Updated
Updated · Ars Technica · Jul 29
AI Engineer Claims Linear A Breakthrough, Assigning 40 Signs and Building 408-Word Lexicon
Updated
Updated · Ars Technica · Jul 29

AI Engineer Claims Linear A Breakthrough, Assigning 40 Signs and Building 408-Word Lexicon

1 articles · Updated · Ars Technica · Jul 29

Summary

  • A self-taught AI engineer said in June 2026 that he had cracked key parts of Linear A, assigning sound values to 40 signs and compiling a 408-word lexicon.
  • The claim began with one proposed Semitic root in a prayer inscription; he then used AI-built scripts to test that sound pattern across the known Linear A corpus.
  • His reading argues that Linear A was an extinct Semitic language, a major departure for a script long treated as a language isolate with no confirmed relatives.
  • The report stresses AI supplied scale rather than the original insight: the engineer made the initial linguistic guess, while the software extended and checked it.
  • Linear A has resisted decipherment for more than a century because it lacks the usual anchors—such as a bilingual text or a clearly related language—that unlocked other ancient scripts.

Insights

Can artificial intelligence truly unlock ancient languages, or is it merely creating statistical illusions from unproven historical guesses?
If a tiny Bronze Age text lacks a translation key, could an AI breakthrough rewrite history or just amplify human bias?