Post #4307276

@cxxvii@aus.social yes, I read about that study too. And I have certainly used LLMs (due to workplace mandate) when it involved more "productivity theatre" than actual productivity. Throughout this year however I've had more and more experiences when it has introduced a step change in productivity on certain tasks. I'm aware that could put me, despite being on the lookout for it, in the category of "thought I was faster, but wasn't". When I experienced big bursts of what I saw as LLM generated productivity, it unsettled me a lot. If I was certain that AI was bad at refactoring, then I wouldn't need to even face all the other ethical dilemmas. The more I become certain that it is in fact quite good at it (with various caveats) the more confused and worried I become. To be clear, I'm not trying to convince you to refactor with AI, just giving context for why I'm bothering to try and use it "ethically". (And to put this in context, I've got 30 years of programming experience, and careful step-by-step refactoring is one of my particular strengths)