Tweet

It seems like everything to do with AI rn has an absurdly wide distribution of outcomes, and the variance of those outcomes is equally massive. Small tweaks to assumptions, or small forks in the path today, lead to wildly different worlds tomorrow. This isn’t isolated to the