What happens when you put AI to work deciphering lost languages? (arstechnica.com)

In a fascinating development for the AI and linguistics communities, a self-taught AI engineer has reportedly made strides in deciphering Linear A, an ancient language with no known relatives and a long-standing enigma for linguists. Utilizing AI-driven programming scripts, he hypothesized that an unknown word in a Linear A prayer inscription could be traced back to a Semitic root meaning "to dwell." His innovative approach allowed him to correlate sound patterns across a collected corpus, leading to the assignment of values to 40 different signs. This effort culminated in a 408-word lexicon, suggesting that Linear A may belong to the Semitic language family, a claim that could reshape our understanding of this elusive writing system. This case highlights the growing intersection of AI and humanities research, illustrating how machine learning can enhance human insight rather than replace it. While AI proved invaluable in testing hypotheses and processing vast amounts of data, the initial creative leap came from the engineer himself. This breakthrough underlines the potential for AI in solving complex linguistic challenges, paving the way for future breakthroughs in understanding not just Linear A but possibly other ancient languages like Etruscan, which also remains partially deciphered.