Linux

The funny thing is that this is the ONE usecase where LLMs can't themselves make anything worse, because if used sensibly it's basically just a black box that pokes at your software until something breaks, the whole point is to break things!

But of course AI-users can't even fucking handle that, they're so utterly incapable of doing any work themselves that "look at the thing and write down things that go wrong and why it went wrong, and check if it has already been reported before submitting your own report" is too much for them to handle.