AI coding tools are optimising the wrong thing and nobody wants to hear it. Writing code was already fast. The bottleneck is everything else: unclear requirements, review queues, terrified deploy cultures, and an org chart that needs six meetings to decide what colour the button should be.
Yeah, I share your unease. There have been a few times where I’ve gotten this vibe from some writing and later found out that it likely isn’t LLM generated text, but it’s always striking to me how this doesn’t ease that uncomfortable feeling — because the thing I’m actually uncomfortable about is how the prevalence of slop has made me so paranoid.
If I’m hyper vigilant about avoiding spending energy reading synthetic text, then I risk unduly dismissing something that someone put real time and energy into writing. But if I’m not cautious enough, I risk wasting my own time and energy engaging with content I’d rather ignore. It sucks to be forced into this position