Re:Dog ejected from car during Sunday crash found on sheep f...
by wizzwizz4 on Thursday June 10, 2021 @04:07PM
(#27465127)
> Your model is busted - the Flynn effect has already reversed and intelligence is decreasing again.
Which metric are we using to measure intelligence this time? How do you know it gives meaningful results for what you're trying to use it to measure? (Please don't say IQ.) And has the study been replicated?
> The risk has nothing to do with what the concrete pressures are, but how much error correction bandwidth we have vs background mutation load rate.
But the selection effects you've described are all local and weak. Worst-case scenario – absolute worst-case scenario – the city Americans marry rural Africans and the problem goes away. But it won't get to that point.
Eliezer Yudkowsky knows a lot about philosophy of thought, and AI theory, but he hasn't put as much effort into learning about genetics. If the average geneticist isn't worried, chances are that he's made a mistake in his reasoning (or that you're misunderstood his concerns) – and I haven't heard anybody in a biological field worrying about this.
---
Wait, are you talking about his “in the absence of selection pressure, we'll get mutations that aren't filtered out”? If so… that's a hundreds of generations kind of problem, and I'm confident that stuff like IVF embryo selection (selecting against the genes for known genetic conditions) is already counteracting that effect. (Yes, this is technically eugenics, yes it could get dystopian if we're not careful, no it doesn't justify what the word “eugenics” makes you think of, no I shouldn't feel the need to write this parenthetical, yes this is the internet we're talking about.)
Transposons are likely to be a bigger issue. I read that they've started duplicating themselves more recently, and we don't really know why.
But imagine the human genome (for some unidentified reason that hasn't happened in the last thousand years) starts degrading across the entire world, so we only have a hundred years left before no human fœtuses are viable. We could just GM a human embryo to be the genetic child of modern-day humans (whose genome we've already sequenced), and we roll-back any mysterious genetic problems that far-future humans might develop.
There is an upper bound to how bad things can get, even imagining something out of a sci-fi novel, so I am highly suspicious of the motives of people who get loudly concerned about the “decline of the human genome”.
Which metric are we using to measure intelligence this time? How do you know it gives meaningful results for what you're trying to use it to measure? (Please don't say IQ.) And has the study been replicated?
> The risk has nothing to do with what the concrete pressures are, but how much error correction bandwidth we have vs background mutation load rate.
But the selection effects you've described are all local and weak. Worst-case scenario – absolute worst-case scenario – the city Americans marry rural Africans and the problem goes away. But it won't get to that point.
Eliezer Yudkowsky knows a lot about philosophy of thought, and AI theory, but he hasn't put as much effort into learning about genetics. If the average geneticist isn't worried, chances are that he's made a mistake in his reasoning (or that you're misunderstood his concerns) – and I haven't heard anybody in a biological field worrying about this.
---
Wait, are you talking about his “in the absence of selection pressure, we'll get mutations that aren't filtered out”? If so… that's a hundreds of generations kind of problem, and I'm confident that stuff like IVF embryo selection (selecting against the genes for known genetic conditions) is already counteracting that effect. (Yes, this is technically eugenics, yes it could get dystopian if we're not careful, no it doesn't justify what the word “eugenics” makes you think of, no I shouldn't feel the need to write this parenthetical, yes this is the internet we're talking about.)
Transposons are likely to be a bigger issue. I read that they've started duplicating themselves more recently, and we don't really know why.
But imagine the human genome (for some unidentified reason that hasn't happened in the last thousand years) starts degrading across the entire world, so we only have a hundred years left before no human fœtuses are viable. We could just GM a human embryo to be the genetic child of modern-day humans (whose genome we've already sequenced), and we roll-back any mysterious genetic problems that far-future humans might develop.
There is an upper bound to how bad things can get, even imagining something out of a sci-fi novel, so I am highly suspicious of the motives of people who get loudly concerned about the “decline of the human genome”.