Artificial intelligence tackles the puzzle of ancient lost languages

Ancient language decipherment has traditionally required an anchor—either a bilingual text or a known linguistic relative. Linear A, the Bronze Age writing system of Minoan Crete, offers neither. Classified as a language isolate with no confirmed connections to any living or dead language, Linear A has challenged linguists for over a century. Etruscan, the pre-Roman language of Italy, fared somewhat better; burial inscriptions have yielded fragments of its vocabulary, yet its grammar and deeper structures remain elusive. These stubborn linguistic puzzles now attract artificial intelligence as a potential tool to overcome centuries of academic obstacles.