Voynich an AI

[Robot59]

Hello Folks,

This is my first post!

The Voynich MS is still not translated after so many years.  Maybe human can't resolve this enigma. Seeing the performances of the robot Sophia (creepy)  and other AI designs which comes all together in the cloud and learn from each other, I think it's possible in the near future with AI. What do you think about this option?

[bi3mw]

Hello Robot59, welcome to the forum.
I think the main problem is that the artificial as well as the "common" intelligence has to know what to look for. There have already been AI solution attempts (see Kondrak and Hauer).
The results have been rather modest so far.

You are not allowed to view links. Register or Login to view.

[Koen G]

My understanding of AI is that it is especially good in working towards a predetermined goal and in the ideal case becoming as good or better than humans in doing that particular thing.

With the VM, we don't know what this goal is. As bi3mw says, we simply don't know which orders to give the AI.

I'm not saying it's impossible, but it's not something the current generation can handle.

[RobGea]

Current AI could make a good transcription. Probably not high on Googles to-do list though,
which is a pity if you think about all those untranscibed mss especially the Oxyrhyncus Papyri.

[Robot59]

Yes, Kondrak an Hauer. Maybe it's an forgotten ancient world language.

[Linda]

Hi Robot59

Welcome

AI is only going to be as good as its programming, both in how it thinks and the data it includes. So you would have to load it up with all known world knowledge (obviously we cannot include that which has been lost) and all possible precedents in terms of literature and imagery, architecture, culture, languages, etc. and even then i don't know if it would work it out since it appears the thinking of the creator(s) of the vms seems extremely creative and different from the norm. 

We can't even yet agree on transcriptions, much less the translation or decryption thereof. I think the diagrams include mnemonics and are therefore not what they seem either. I don't think machine-based intelligence is going to be much of a help given these issues.

[Koen G]

Current AI could make a good transcription. Probably not high on Googles to-do list though,
which is a pity if you think about all those untranscibed mss especially the Oxyrhyncus Papyri.

I've been thinking this as well, they should train AI to transcribe manuscripts. This is something which probably could be done with enough money and dedication. And it would mean a huge advancement in various studies. Won't help directly with VM though.

[Anton]

I've been thinking this as well, they should train AI to transcribe manuscripts.

There's still considerable room for improvement in OCR of printed texts...

[davidjackson]

I've been thinking this as well, they should train AI to transcribe manuscripts. This is something which probably could be done with enough money and dedication. And it would mean a huge advancement in various studies.

Sadly, no.
You have to train an AI on a very specific dataset. Which means that, if you want to transcribe a manuscript, you have to teach the AI what each individual letter represents.
In the case of the VM, it would be impossible, because from the very start, your personal preferences would guide the software. Two different people would get two different results...

[-JKP-]

The world of "AI" is not as "I" as people may think.

The creators of the Google Go-playing software AlphaGo admitted that there wasn't a lot of AI in it, that it's mostly brute force... a lot of capturing of games and consequent databasing of that information, with lookups of the kinds of moves that pros make in those situations.

It can be useful. For example, in visual imagery, software can be trained to make better guesses, but it's almost always with a human sitting there with the software saying, "No, that's not a good guess, try something different. Yes, that's a better guess, try more of that kind..."


I'm particularly interested in AI—it's what I was studying in my graduate work (AI as it relates to the analysis and generation of visual imagery), and I think there's fantastic potential in a number of directions, but MUCH of the AI that exists today still takes its data from huge databases (without knowing the rationale for why people do things in a certain intelligence-informed way), and humans still have to kind of point the software in the best directions.