@bkase_@leonardoalt@bendlang Most proof languages are also a programming languages as you need a way to define functions, and as soon as you have that you basically have a programming language.
Nice cool work from people at MIT, using Rocq to verify operating systems and tools around, starting with xv6, a Unix system designed for teaching: https://t.co/VVW8BjyIJ4 ๐ค
@florian_tramer An idea: the programs are actually not totally random, as a human (or LLM) could figure it out with mostly no bugs, so there must be a way to prove it right without combinatorial explosion, taking the right path.
If AI review can find any bugs, why are there still tests in apps? ๐ค
This remark should show why AI alone is not enough for code review. Unless you have other though?
This is going to sound dramatic but I am genuinely starting to think that weโre going to see pre-AI software as a kind of โhow did they even do itโ endeavor. People will look at Linux like the pyramids