leogao comments on Vote on Interesting Disagreements

[leogao]

An alignment technique that can fully align GPT-4 is likely (>50%) to also fully align the first existentially dangerous AGI