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 Second, already out of date - Okbay et al 2016 has increased the number of genome-wide significant education hits to 162: https://www.dropbox.com/s/my9719yd8s5hplf/2016-okbay-2.pdf Third,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...
My 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.
This 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).
We 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.
> 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.)
As soon as the odds approach 40-50% that the traits I select would be expressed, I'm willing to start spending tens of thousands of dollars. Its an investment in the future, with probably the biggest return I could provide my child(ren).
I think you're misunderstanding. Traits like 'intelligence' are always 'expressed'. They aren't binary and ~100% determined by genetics like gender. You don't select for 'would be expressed', you select for something like 'on average, has 0.1SD higher measured trait values'.
Even a trait which seems to be binary, like schizophrenia, is actually probably on a continuum (relatives of schizophrenics have lower IQs, many people have symptoms which wouldn't rise to the level of a full diagnosis) and can be seen as a liability threshold model where a large number of genes each have deleterious effects, yielding a normal distribution of risk and the higher your risk, the more likely random chance or an environmental insult will push you into a full-fledged schizophrenia. So if you were selecting on mental illness risk, it would be impossible to select for someone who 'does' or 'does not express' schizophrenia, but it would be perfectly meaningful to talk about halving or doubling the risk, which would translate to 0.5% vs 2% (in almost all of the people selected for less schizophrenia risk, they wouldn't have had schizophrenia either way; what it buys you is 1 less future out of 200 futures in which the person becomes schizophrenic).
> 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.
> 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 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!)
Can't wait to be able to clone myself... well probably won't happen for me personally. But sexual reproduction is so annoying. You have to have another person to deal with, and you don't know what the hell you are going to get out of the dael.
It's great if you can find that mythical partner. The problem is finding that partner, especially if you're not very attractive or not very social.
Hopefully, in the future we can fix the former (current cosmetic surgeries are just barbaric hackery, I mean at the cellular level, like rewriting someone's genome to make them taller and more attractive), and then extend lifespans so people have more time to find good partners instead of having to settle.
Aside from Sophistifunk's answer which is a pretty good one, replace your adblocker with uBlock Origin and subscribe to the lists: "Anti-Adblock Killer | Reek" and "Adblock Warning Removal List".
I remember that either "EasyPrivacy" or "Fanboy's Enhanced Tracking List" helped kill some that the other two lists didn't, so I just subscribe to both.
To the contrary I think there's a decent argument to have about the idea that if we cloned a few billion people into being, it could help solve our problems faster, through their brain power and the increase in competition. There'd be less to eat, and more short-term pollution, but it might be beneficial long-term, since most of the CO2 to have been released has already been released cumulatively during the last century.
We can also arrive at Brave New World style human production. Babies are produced in factories according to characteristics projected to be in demand by the time they have fully grown. Except instead of the alpha-beta-epsilon types in the book, we may be able to breed for maximum diversity for maximum potential resistance against diseases. Or, of course, any other objectives, technology permitting.
We may also arrive at a common trope in sci-fi where reproduction is not an individual (or couple) decision. By default, individuals do not reproduce and the factories keep the society populated.
As nice as the sentiment sounds, a majority of our most brilliant people now are working on improving click-rates and keeping the stock market efficient. I don't really have hopes for the utopia you speak of.
Younger people are already spurning a lot of careers that aren't "fulfilling". We absolutely could handle a huge increase in human population, if we did the following:
- institute Universal Basic Income
- massively increased the density in cities and built more housing in them (i.e. more skyscrapers, and link them together better so cities are more 3-dimensional)
- greatly improve local transportation using PRT (personal rapid transit) like SkyTran for less-dense areas and more-expensive service and better mass-transit systems for the highest-density areas
- improving long-distance transportation with Hyperloop, high-speed rail (maglev), maybe vactrains later, etc.
- more efficient farming, such as the "vertical farming" techniques being explored now
- more automation of farming and other industrial production (which feeds back into making UBI work)
- cleaner energy sources, mainly better utilization of solar energy with increased PV and maybe orbital solar collectors
- asteroid/moon mining and maybe moving dirty industries offworld (with energy being clean and renewable, I'm thinking that metal refining will probably be the last really dirty industry, and with more metal mining offworld, it'd make more sense to do the refining there too)
Perhaps the main cultural benefit will be more people actually planning their reproduction and families. Of course this will be shortly followed by a "natural born" rights movement, where the unengineered (but virtue of being less competitive) demand to be recognized as a protected class.
Good luck telling the "natural born" apart from the engineered. Obviously, the more extreme the engineering, the easier it'll be to detect, but more subtle tweaks will not.
All this "designer baby" stuff sure will take the fun out of unprotected sex though.
Excellent article and a nice balance between discussing the technology and the ethical implications. As a technologist I find it frustrating that the media almost always focuses on the ethics.
While the politicians in certain locations might seek to constrain this type of research, often on religious grounds, it will simply move offshore to more liberal regimes. As the technology progresses, the wealthy will utilise it to engineer "designer babies". The politicians and ethicists may as well accept that this research and development is going to happen; they can simply choose whether they want to control and benefit from it or give away the lead.
34 comments
[ 2.5 ms ] story [ 44.7 ms ] thread> 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 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...
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.
(great film, if any of you haven't seen it)
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.
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.)
Even a trait which seems to be binary, like schizophrenia, is actually probably on a continuum (relatives of schizophrenics have lower IQs, many people have symptoms which wouldn't rise to the level of a full diagnosis) and can be seen as a liability threshold model where a large number of genes each have deleterious effects, yielding a normal distribution of risk and the higher your risk, the more likely random chance or an environmental insult will push you into a full-fledged schizophrenia. So if you were selecting on mental illness risk, it would be impossible to select for someone who 'does' or 'does not express' schizophrenia, but it would be perfectly meaningful to talk about halving or doubling the risk, which would translate to 0.5% vs 2% (in almost all of the people selected for less schizophrenia risk, they wouldn't have had schizophrenia either way; what it buys you is 1 less future out of 200 futures in which the person becomes schizophrenic).
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.
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 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!)
Hopefully, in the future we can fix the former (current cosmetic surgeries are just barbaric hackery, I mean at the cellular level, like rewriting someone's genome to make them taller and more attractive), and then extend lifespans so people have more time to find good partners instead of having to settle.
In any case, all your ancestors managed to procreate (duh!). So you can't blame your genes too much.
Also, ad-blocker-blocker.
I remember that either "EasyPrivacy" or "Fanboy's Enhanced Tracking List" helped kill some that the other two lists didn't, so I just subscribe to both.
We may also arrive at a common trope in sci-fi where reproduction is not an individual (or couple) decision. By default, individuals do not reproduce and the factories keep the society populated.
Efficient allocation of capital is a noble endeavour.
If you select half of the parents material and then just double it, that's sounds like it would be maximally inbred.
If you just select all of it, that's a clone.
All this "designer baby" stuff sure will take the fun out of unprotected sex though.
While the politicians in certain locations might seek to constrain this type of research, often on religious grounds, it will simply move offshore to more liberal regimes. As the technology progresses, the wealthy will utilise it to engineer "designer babies". The politicians and ethicists may as well accept that this research and development is going to happen; they can simply choose whether they want to control and benefit from it or give away the lead.