So why can't AI help in solving the last Zodiac cipher?Could it be that it's just a dumb autocomplete on steroids hyped by marketing?
Maybe it is just jibberish and has no solution.
>>109400470>If the AI cannot solve a problem then it's just a dumb stochastic parrot>If hudreds of human experts cannot solve a problem then... IT'S THE UNBREAKABLE SPIRIT OF HUMAN SACRIFICE AND VITALITY OK???
>>109400486The first two had solutions though.
>>109400521>the first two have solutions, so the third one definitely has to have a solution!Have you ever wanted to troll people?
>>109400536It's the fourth one, the third one is likely the troll one but that one got solved though it's too short to be uniquely verified. My theory is that it was solved correctly but that the Zodiac was clever enough to offset the blame.
>>109400521how did they find the solutions?
>>109400506It's called computational offloading, I guess we need to solve everything by hand without calculators using log tricks.
>>109400562First one was a simple homophonic substitution cipher, the second one was the same with an added knight move (chess) transposition. It was solved in 2020.The third one got solved by brute force narrowing and substitution on a qualitative background. As in narrow down names in the area at that time with potential motive.
>>109400620>bringing up homos out of nowheremindbroken
Didn't they make distributed computing @home projects for other ciphers? If they made one for this I'd donate my compute.
>>109400620I thought it all'd be the same kind of cypher, interesting
>>109400565It's called double standard
>>109400636Of all things to make me laugh
>>109400636Lol, good thing I didn't show you my homomorphic mapping function I use in CV.
>>109400620very interesting.do you think the people who deciphered them are above 100IQ geniuses?>>109400636thanks for the laugh bro
>>109400470The third cypher had a mistake in it that caused it to go unsolved for much longer than it would have if it didn't. The guy was a fucking idiot not the super genius modern media wants to make him out to be. It's very well possible that even if the final one isn't a straight up troll it's so riddled with mistakes as to be meaningless because to solve it you would have to correct that midwits fucked up cypher first.
>>109400636I hate that this made me laugh.
>>109400932He accidentally skipped a single character space while copying his transposition onto the final page.It can happen to anyone in a hurry to get the material out. But it's interesting how one small mistake threw off computers looking for perfect patterns. I thought LLMs utilize fuzzy logic?
>>109400888The first got deciphered by a wife and husband at home. The second and third one did require a multi year effort by pros though the second one had a transpositional error inflating the difficulty.
>>109400470It rather looks like human intelligence is just dumb autocomplete on steroids, too! It obviously doesn't mean everyone can solve these ciphers, some of which may even be faulty.
>>109400985The problem isn't getting a solution but proving it.If you look for it every year there's at least a dozen research teams claiming to have solved the final zodiac cypher, K4, etc.What AI has helped establish is for example that the Voynich manuscript definitely is a real language mathematically speaking, but it doesn't get us one bit closer to actually reading the script, we just know yeah it's almost certainly real since the alternative is an unbelievably complex hoax that perfectly follows established linguistic analysis patterns across all human languages like distribution, combination odds, etc.Similar for the Indus Valley script, we now know yeah it's real, it's not just a billboard or advert or sigil or family crest, it is part of a language, but we still haven't conclusively deciphered it.When you give a sufficiently advanced AI model (be it an LLM or a NLP model specifically trained on pictographic languages) an undeciphered script it can spit out reasonably possible translations (as in actually consistent translations across the data we have and the sentences do make sense) but that doesn't prove it's real. It's just a possible solution.These things don't have a Rosetta stone that makes it easy to prove what the correct solution is because you already have it in a different language, the Voynich manuscript, Indus Valley language, etc are too isolated to easily prove your translation is correct and not one of the 2347982374 other proposed ones.
>>109401221>What AI has helped establish is for example that the Voynich manuscript definitely is a real language mathematically speakingThat's interesting AF, I was sure it's fake.
>>109401142>It rather looks like human intelligence is just dumb autocomplete on steroids, too!Human intelligence is rooted in the Nous or rather linked to it. It's uniquely linked to humans.
>>109401259I was always 50/50 on it being fake or a really elaborate obfuscation tactic and it's looking like it's not a fake. It's not impossible but it would require the author to have an insanely advanced understanding of linguistics that they built into a hoax they kept so consistent they made intentional mistakes that they then corrected themselves (e.g. using the wrong "translation" crossing it out and using the correct one next to the crossed out one).Looks like it's a medical pseudo-religious journal something similar to what you'd expect druids to have written down if they weren't an oral tradition.I.e. it was most likely written by some guy that had some advanced medical knowledge he wanted to hide for whatever reason (religious persecution or simply just job security) mixed with mythologizing (the seemingly fertility related bath section) and medicinal instructions (the weird plants are likely instructions what parts of which plants to mix in what measure) but we still don't have a conclusive translation. All of this is just established consensus (read speculation) based on the fact that we now know yeah it's real and the graphics in the manuscript itself.
>>109401904I'm going down a deep rabbit hole on it now, it's super interesting. But as you said, it has no reference point so who or what can really say which theory is the correct one, can AI aid in this at all?Likely not.
>>109404522Not really. I might be wrong on this but as I recall basically all major linguistic breakthroughs in decoding dead languages happened because we found some kind of Rosetta stone that showed us conclusive what the translation should be.The Voynich manuscript for all we know is unique, it does have ~30k words (again iirc) but that isn't enough to do anything more than fancy bruteforcing.I'm not totally up to date on the Voynich one specifically but I'm willing to bet there are probably research teams out there that have come up with plausible translations using some kind of AI model whether it's a LLM, NLP algo, or something else, but proving it right might actually be impossible.The Indus Valley one might be even more hopeless because we barely even have full sentences for that one, just lots of fragments.
>>109400470Its a homophonic substitution encryption and the cipher text is too small to make an assumption about the key or frequency of the symbol space.It could try known keys or transpositions of known keys but otherwise you're mathematically shit out of luck if it's a unique key.
