Calgary engineer believes he's cracked the mysterious Voynich Manuscript

[Koen G]

For my last video, I asked ChatGPT to come up with a sentence and then encode it using simple substitution. It will spit out a sentence and a cipher alphabet, then encrypt the sentence and make a bunch of mistakes. When confronted with the mistakes, it will correct them and instead make other mistakes. I ended up not using this exercise for the video, because it was useless. 

Current generation AI is not smart. It is simply good at producing convincing language. Do not trust it for complex analysis! By using ChatGPT for this kind of task, all you are doing is signaling your lack of understanding of its limitations.