5 ms·
Bizarre that both Greely and Ball seem to be unaware of iterated embryo selection, which is far more mind-boggling than anything else covered in the article. >
by gwern 11y ago
Bizarre that both Greely and Ball seem to be unaware of iterated embryo selection, which is far more mind-boggling than anything else covered in the article.
> Intelligence runs in families, but there is no guarantee you’ll get a good dose. While putative “intelligence genes” have been sought for years, so far all we have found is a handful of genes that collectively account for less than two IQ points. Greely thinks this will improve as we get more adept at gene-hunting. I am not so sure.
Several misunderstandings here. First, it is unimportant whether you have a few or many SNPs which pass a thoroughly arbitrary threshold of p-values. P-values are not posterior probabilities are not effect sizes are not utilities are not profits are not decisions. What matters here is the fraction of variance the full polygenic score can explain. Second, already out of date - Okbay et al 2016 has increased the number of genome-wide significant education hits to 162: https://www.dropbox.com/s/my9719yd8s5hplf/2016-okbay-2.pdf https://www.dropbox.com/s/my9719yd8s5hplf/2016-okbay-2.pdf Third, this will go up monotonically as sample sizes increase, as it has already gone up as sample sizes increased, as has happened with other traits where GCTAs confirmed substantial SNP heritability. Fourth, one selection step may be unimpressive, but many will be much more impressive, bringing us back to iterated embryo selection, which is the true importance of the mentioned breakthroughs in gamete<->stem-cell conversion techniques...
- toomuchtodo 11y agoMy wife and I went through several IVF rounds, and our doctor even indicated that we could have our embryos genotyped to select for certain traits. We didn't, but we could have. The phrase, "Their doing some very interested research in China right now" was said during one of our consultations. I have no illusions that in 10-20 years, I could have an embryo not only selected for traits I'm interested in, but the embryo edited using CRISPR if necessary to express the traits wanted if not already present. I wouldn't even need reproductive tissue to do this. Interesting times. It feels like biology/genetics is on the verge of exploding similar to what happened to silicon in the 70s, with far greater consequences for humanity.
- hkmurakami 11y agoBrb rewatching GATTACA. (great film, if any of you haven't seen it)
- nickpsecurity 11y agoI second that for anyone considering these things.
- Grishnakh 11y agoThis is rather scary, and reminds me a of a 90s episode of The Outer Limits where all the middle-aged people are losing their jobs because they're not as smart as the 20-somethings who are all genetically modified (even though it's illegal).
- gwern 11y agoDid you ask what traits were on offer? What do you think it would have taken for you to taken the offer?
- toomuchtodo 11y agoWe were told the easiest option was to select gender. Other traits were going to be harder/more expensive/uncertain outcomes. Intelligence. I would've selected/edited for intelligence if it had been a sure thing. Life is hard, and while I don't care much for superficial traits (hair/eye color, height), I want to metaphorically lift my child as high as I can against his/her peers.
- gwern 11y ago> Intelligence. I would've selected/edited for intelligence if it had been a sure thing. Gender is special in being easily chosen with ~100% accuracy, either via embryo selection or simply selective abortion. But for everything else, it's more a question of reducing the odds or increasing the average. How much of an average increase in intelligence would you want before you did it? > Life is hard, and while I don't care much for superficial traits (hair/eye color, height) You know all of those correlate strongly with career success and income, right? Taller better-looking people get paid substantially more. (And the Mendelian randomization on BMI, IIRC, indicate it's causal.)
- visarga 11y ago> one selection step may be unimpressive, but many will be much more impressive, bringing us back to iterated embryo selection, which is the true importance of the mentioned breakthroughs in gamete<->stem-cell conversion techniques... I saw a video recently that said if we can select 1 in 10 embryos, we can gain 4 IQ points. If selection is 1:1000, we gain 20 or more IQ points. This process can be repeated between embryos, so there is no need to wait 20 years and convince them. Also, they said that when we have completely scanned a few millions of people's DNA we will have the data to discover more complex dependencies. The company in question wants to scan 10 million people in a decade. By the time we reach an AI singularity, we might have evolved biologically as well. Maybe we could grow bigger or more complex brains for us and be just as good as the AIs.
- gwern 11y ago> I saw a video recently that said if we can select 1 in 10 embryos, we can gain 4 IQ points. If selection is 1:1000, we gain 20 or more IQ points. This process can be repeated between embryos, so there is no need to wait 20 years and convince them. Something like that. It sounds like that video is echoing Bostrom & Shulman's paper. The possible IQ gains depends on whether you're assuming you have a polygenic score which explains all of SNP additive heritability of intelligence (and how much you estimate that at) and whether you take into account the losses in the IVF process which mean that the best-scoring embryo will probably not yield a live birth. If you are interested in the details, I've written it all up and provided R code so you can simulate out various scenarios in my http://www.gwern.net/Embryo%20selection http://www.gwern.net/Embryo%20selection page. My best guess is that a realistic gain at the moment is closer to 0.4 IQ points than 4 IQ points, but iterated embryo selection could be a huge gamechanger and I'm really surprised that people aren't taking it more seriously. Embryo selection is the present; CRISPR is the next 5-10 years; but iterated embryo editing/selection may be everything after that... > Also, they said that when we have completely scanned a few millions of people's DNA we will have the data to discover more complex dependencies. The company in question wants to scan 10 million people in a decade. Yes, I'm sure. I think Venter's company might be the one in question, but it's focused on rare diseases, and may not be collecting psychological phenotype data like IQ or personality. (Hard to run a GWAS for intelligence if you have a million genomes without any associated data like IQ test scores or educational credentials!)