I sort of agree, if you can guarantee the AI generated proof isn't a false positive. To be fair though, your example would be easy to explain to someone. You ju

I sort of agree, if you can guarantee the AI generated proof isn't a false positive. To be fair though, your example would be easy to explain to someone. You just show how the encryption algorithm can be trivially broken. A program that can break major encryption algorithm would likely be understandable, or at the very least we could show how it can decrypt things.