I'm really appreciating some of the writing en AI that is coming out now. Beyond simply "vibe coding is bad" but advancing arguments about the specific risks.