The Illusion of Agreement: Why ρ = 0.01 Ruins Scalable Oversight
I was reading this paper, and I noticed a massive unaddressed exploit in an open problem (which the authors flagged), this caused me to absolutely lose my mind and stay up until 6:00 AM hyperfixating on the math. After sleeping three hours I woke up at 9:00 AM out of...
Aug 291