There are a number of silly stories out there right now about Artificial Intelligence. None more silly than AI translating the unintelligible Voynich Manuscript (such as the one in News Week). The Voynich Manuscript is from 15th cent. and is basically written in its own language (or possibly a coded version of an existing language). Its meaning has eluded linguists, lexicographers, and cryptographers for over a hundred years. The rabbit hole of an article in Wikipedia on the Voynich Manuscript is a sufficient introduction. The “translation” of the the manuscript sounds a lot like the bad, early mistranslations of Summerian tablets, Egyptian hieroglyphs, or even Ezra Pound’s erstaz Chinese. Here again we have an example of the problem with AI – we provide questionable or incomplete input into a computer and believe that we have an accurate result based on our faith in the algorithms.
I have an Amazon Echo Dot at home that I got from my brother-in-law for Christmas. A simple translation search via Alexa cleared the Voynich translation up very neatly. When I asked her to translate the first sentence she claims it means “I am sorry. I do not have enough information on that.” This is a strange way to begin a book, but there could be some missing pages. It also seems that they used a “404” numbering system. Which either means it is a five-based numbering system or one that is 404 based.
I don’t think those doing textual analysis on the Voynich manuscript believe that they have translated it. If you read the stories, they believe that they have made an advance. But that would not generate traffic would it?