The AI scare is crazy. I understand and agree that there are legit uses of LLMs especially in the context of searching and summarizing which it does an excellent job of. But outside of those domains it isn’t that great. Especially for coding tasks. It creates spaghetti code that’s hard to troubleshoot. SWE is not just about writing code it’s about writing MAINTAINABLE code. You can one shot a feature and it sort of works but inevitably when the user demands change that’s a world of pain for you.