AI May Have Unlocked the Secrets of the Mysterious Voynich Manuscript

This site may earn affiliate commissions from the links on this page. Terms of use.

Cryptographers and linguistic experts have been puzzling over the Voynich manuscript ever since its discovery over 100 years ago. This confounding 240-page codex was written centuries ago in an indecipherable gibberish language that has remained untranslated despite extensive study. Researchers from the University of Alberta now claim to have made significant progress in understanding the manuscript with the aid of artificial intelligence.

The Voynich manuscript is named after the Polish book dealer who uncovered the book in 1912, but little else is known about it. The parchment has been dated to the early 15th century, and experts believe it may have been written in what is today northern Italy. The book contains page after page of strange script, which goes right to left across the page. Many pages also include depictions of plants, astronomical diagrams, and human figures.

Some investigators have suggested that discoverer Wilfrid Voynich may have forged the manuscript using antique materials, but neither Voynich nor the other suggested authors have been confirmed as the true creator of the manuscript.

Greg Kondrak and Bradley Hauer from the University of Alberta aren’t concerned with authorship, but in figuring out what the manuscript says. The AI model developed by the pair first needed to determine what the encoded language of the manuscript was. To do that, the network was trained with the Universal Declaration of Human Rights translated into 380 different languages. When shown the Voynich text, the AI identified it as encoded Hebrew.

Working from the assumption that the manuscript was indeed written in Hebrew with a cipher, the researchers looked at how to transform the text into something readable. One popular hypothesis about the manuscript is that the text is a form of alphabetically ordered anagram (the letters are jumbled). The researchers devised an algorithm to unscramble the characters and turn them back into standard Hebrew characters. They found that around 80 percent of the words in the book were in the Hebrew dictionary.

Armed with this process, Kondrak and Hauer translated the opening passage, but experts on archaic Hebrew couldn’t make heads or tails of it. After some spelling corrections, they came up with something that might be accurate: “She made recommendations to the priest, man of the house and me and people.”

It’s a strange phrase, but this could be a big step toward understanding the manuscript. Linguists are not yet convinced, but translating encoded text from the 15th century is bound to be inexact. Even uncoded texts from that era are difficult to follow today. Time will tell if Kondrak and Hauer are on the right track.

Let’s block ads! (Why?)

ExtremeTechExtreme – ExtremeTech