tailcalled comments on Contra Nora Belrose on Orthogonality Thesis Being Trivial

[tailcalled]

The orthogonality thesis is indeed not sufficient to derive everything of AI safety, but that doesn’t mean it’s trivial.