That’s not contradictory. It’s simply true that average and median competence get higher both over time for people (peaking in the 30s-50s depending on dimension of competence), and get higher as you look at more respected cohorts (Navy SEALs, CEOs, etc.). It’s _ALSO_ true that when compared to naive expectations, ain’t none of us worth a damn. It’s simultaneously true that nobody is ACTUALLY competent, and that some groups are impressively less incompetent than others.
It’s a question of how much of the Y axis you choose to look at. Pretty much all humans are in a very narrow band of capability when put on a gradient from free-floating hydrogen to a theoretical future galaxy-spanning AI. On that absolute scale, even comparing Iain Banks’ Ship minds to flatworms is only a minor gap. On the RELATIVE scale of median modern human to elite (top decile, say) modern human, it seems like a nearly qualitative difference.
Hmm. I read it as “understand that there are people significantly more competent than others”. What have I missed? In his closing, he writes:
But I’m pretty sure that, statistically speaking, there’s a lot more cream at the top than most people seem willing to admit in writing.
He didn’t talk about the other part of my comment (that humans are a very small range of competency spectrum and NO human is very smart compared to imaginary perfect minds). That idea was the taken from the current post, which I tried to show was NOT a counterpoint, but just a compatible observation.
I think you’re right. I had a memory of the post where Eliezer was trying to say something like “We’re all in good hands,” but after rereading I don’t see that as much.
That’s not contradictory. It’s simply true that average and median competence get higher both over time for people (peaking in the 30s-50s depending on dimension of competence), and get higher as you look at more respected cohorts (Navy SEALs, CEOs, etc.). It’s _ALSO_ true that when compared to naive expectations, ain’t none of us worth a damn. It’s simultaneously true that nobody is ACTUALLY competent, and that some groups are impressively less incompetent than others.
It’s a question of how much of the Y axis you choose to look at. Pretty much all humans are in a very narrow band of capability when put on a gradient from free-floating hydrogen to a theoretical future galaxy-spanning AI. On that absolute scale, even comparing Iain Banks’ Ship minds to flatworms is only a minor gap. On the RELATIVE scale of median modern human to elite (top decile, say) modern human, it seems like a nearly qualitative difference.
That doesn’t seem to be what Eliezer was implying though.
Hmm. I read it as “understand that there are people significantly more competent than others”. What have I missed? In his closing, he writes:
He didn’t talk about the other part of my comment (that humans are a very small range of competency spectrum and NO human is very smart compared to imaginary perfect minds). That idea was the taken from the current post, which I tried to show was NOT a counterpoint, but just a compatible observation.
I think you’re right. I had a memory of the post where Eliezer was trying to say something like “We’re all in good hands,” but after rereading I don’t see that as much.