Conspiracies, love letters, and medieval remedies: how artificial intelligence helps decipher the secrets of the past

Artificial intelligence is emerging as a powerful tool for historians and researchers, unlocking secrets hidden within thousands of coded historical documents scattered across global archives and libraries. These texts, some remaining undeciphered for centuries, are now being opened up by AI, offering a new window into lost worlds. One significant breakthrough involves the "Borg cipher," a mysterious manuscript housed in the Vatican Library. For over 400 years, its 408 pages, filled with unknown symbols and difficult-to-interpret text, have resisted decipherment. The document was believed to contain secret medical recipes and treatments. However, international researchers, employing machine learning algorithms, have successfully cracked the code. Discoveries include unusual remedies, such as consuming quality red wine or fermenting nutmeg in dough to treat dysentery. Beáta Megyesi, a professor of computational linguistics at Stockholm University, described the process as akin to detective work, where each symbol and partial solution brings researchers closer to understanding past lives and a vanished historical world. It is estimated that approximately 1% of archived documents are fully or partially encrypted. These hidden texts contain a diverse range of information, from diplomatic correspondence and secret society rituals to medical treatises, love letters, and details of daily life that authors intended to keep private. Deciphering them can significantly alter the understanding of historical figures and entire eras. For instance, an analysis of encrypted letters attributed to Mary, Queen of Scots, revealed her involvement in plots to regain the throne and her strained relationship with her son, the future King James I of England. Cracking these old codes presents varying levels of difficulty. Some ciphers, like the Borg manuscript, use simple substitutions where symbols correspond to Latin letters. Others are far more complex, with researchers sometimes unaware of the original language. Some documents intentionally include misleading symbols to deter unauthorized readers, while others use multiple different signs for the same letter. These complexities transform the decipherment process into a large-scale puzzle. Criptolog Cecile Pierrot, for example, spent six months deciphering a 500-year-old letter from Emperor Charles V.

“

It is like detective work. Each symbol, each pattern, and each partial solution brings us closer to the secrets of people and to a vanished historical world.