Notes on countermeasures for exploration hacking (aka sandbagging) — LessWrong
lesswrong.com · saved by 1 readers
If we naively apply RL to a scheming AI, the AI may be able to systematically get low reward/performance while simultaneously not having this behavio…