habryka comments on Six Thoughts on AI Safety

[habryka]

I don’t currently think it’s plausible, FWIW! Agree that there are probably substantially easier and less well-specified paths.