Cultural drift is a big problem if selection pressures are the only way to correct errors. But is that the case? Evolving the right culture used to be the only way to survive as a group, but now we have science that produces GMO crops and mRNA vaccines. We don't rely on culture to solve food production or surviving plagues anymore, at least not to the same degree. We don't need our culture to die in famines and pandemics in order for another culture to replace us that happens to value bioengineering and social distancing. Similarly, couldn't problems caused by cultural drift be corrected in ways other than cultural selection?
Granted, some problems might be too subtle or complex for anyone to understand their consequences, or convince enough people of their importance. Especially if the problems affect our ability to understand and correct problems. Fertility is such a problem (if low fertility halts innovation), but we're aware of it, have some ability to intelligently control our collective behavior, and therefore we might solve that through means other than cultural selection.
In many ways, cultural evolution took over from genetic evolution, when genes for culture evolved. Could we be (or become - through futarchy or other institutions which improve on what's there) a culture which invents a mechanism that alleviates cultural drift?
If I understand right, a rational culture searches for and adopts values that are adaptive. I'm not sure how far you propose to take this. What I had in mind was to mostly keep our current values, even if they aren't the most adaptive ones, and even to allow some drift in values (as long as we don't fall away from core Enlightenment values). I don't like the prospect of low fertility not because high fertility is adaptive, but because I like innovation, freedom and tolerance, which are current values I inherit from my culture. If there's a way to fix problems without compromising current values, or by compromising less sacred values in order to safeguard more sacred values, I'm for that.
How far into the future do your values extend? The longer you want us to last, the more you'll have to attend to adaptation. Non-adaptive cultures don't last.
I imagine there's a tradeoff between being adaptive and keeping our current values, but also that there could be quite a lot we can do to be more adaptive without significantly compromising our values. Perhaps values like freedom and tolerance are not the most adaptive overall, but they are quite adaptive - if we're talking about anything like human psychology, these values could constitute an attractor state, simply because they appeal to individuals. Maybe as long as we stay in the game to provide the example, more fertile (or otherwise more superficially adaptive) cultures eventually adopt our core values, because they want to.
But if not, and it's a certainty that a set of values without freedom, tolerance and other nice things eventually outcompetes us, I'd say it's better to have lasted for a while with something like our values, than to last longer with bad values.
“as long as we don't fall away from core Enlightenment values“
Why?
Core Enlightenment values found their ultimate expression in the drownings at Nantes, the Reign of Terror, and several subsequent iterations of modernist totalitarian regimes.
Are you really so wedded to those values at all costs?
Killing in the name of peace does not make killing an expression of valuing peace. So I reject that your examples are expressions of enlightenment values.
If the French revolutionaries were committed to the Enlightenment value of using reason over violence, they would have reasoned with their political enemies instead of murdering them.
1) Longevity technology - doesn't add more humans, but slows down (or halts) dying, so in the long run results in a positive population growth.
2) Anti-aging/more effective IVFs - allows having children over a longer period. Or, on the other hand - more widespread IVFs, resulting in a higher proportion of pregnancies being twins.
3) Changing cultural norms around surrogacy - already accepted to some extent, if widespread then fertility might then be more exposed to market incentives and thus direct govt control
4) Artificial wombs - the solution I'd be most bullish on, allows countries to unilaterally decide on setting up a human export operation. Also allows men to unilaterally make a decision.
5) More advanced genetic testing, lowering the risk of having a child (a frequent argument against - a huge risk the the kid needs a lot of care, and no way of hedging it). Related: changing cultural norms around abandoning sick children/putting them under state-sponsored care.
For one, the reign of terror happened not because people were committed to using reason, but because they chose violence and abandoned reason. If they had been committed to reason, they would have reasoned with their political opponents instead of murdering them. Just because they said they did it for liberty and equality doesn't mean it's true.
What's with the "Malthus"? Malthus is much criticised in modern times. He famously argued that reproduction would tend to outstrip resources. However, his argument used bad math - he argued that resources could only increase linearly. Are we supposed to forget about that second part - while giving him credit for the first part?
The concept of "cultural drift" seems pretty similar to what's usually called "evolutionary mismatch". Both often involve lists of maladaptive traits that used to be adaptive.
Pronatalist intellectuals should focus on aiding totalitarian regimes like China, Russia, and North Korea in addressing demographic collapse. Success there could pressure Western governments to pursue reforms.
You have written "I’ve theorized that an essential trend of the modern world is “cultural drift”, i.e., selection becoming much weaker relative to internal drift in driving/disciplining the cultural features that are harder to vary within cultures."
This seems rather fuzzy to me. The link you provide is paywalled and I did not see an accessible version on this substack. But I seem to remember reading some of what you wrote on cultural drift in the past (was the Quillette piece not paywalled in the past?) that led me to think you were using cultural drift in its older sense (see link) and not as the cultural analog of genetic drift, which only happens in small populations and would not be active in societies as large as ours today. But I recall you writing favorably about Boyd and Richerson, whose cultural evolution system I use. They don't see falling birth rates as cultural drift but rather to something more akin to the peacock's tail, IIRC.
They see falling fertility as due to dysfunctional status markers. And I'd say those markers drifted into dysfunction, in that selection didn't keep them on an adaptive track. By "drift" I mean change not disciplined by selection.
They are not dysfunctional in a cultural evolutionary sense. They are in a gene evolutionary sense. This makes them sort of like peacock tails that make make peacocks less able to survive, but more attractive (fit) as sexual partners.
It is not clear that declining fertility is the problem some make it out to be. Medieval cities experienced negative natural increase, only maintaining their population and even growing by in-migration form the countryside. Excess agrarian population moved to the cities where they pursued nonagrarian occupations. Most of the innovation happened there. In the future we could see the developed regions of the world playing the role of cities with the undeveloped world playing the role of the countryside.
Alternately if we want to boost fertility at home we could try abandoning neoliberalism and returning to the stakeholder capitalism of my youth. That's what I would recommend.
Genetic drift still happens in large populations. Consider that most of the human genome is typically junk DNA. In such regions, genetic drift is unconstrained. Indeed, the molecular clock relies on such genetic drift.
My understanding is genetic drift usually refers to the loss of rare, but adaptive genes do to chance. This generally only happens in small populations. Mutations in junk DNA doesn't have the same effect.
Cultural drift refers to the loss of cultural information due to random chance. The example I've seen given is Chatham island peoples, who when they first arrives had the full tool set of mainlanders, but gradually lost much of what they once had as rarely used cultural elements were lost simply because a time came when no one was taught these rarely used skills and so they were lost. In a big population there will always be a handful who learn rarely used lore who carry the knowledge on and then enjoy a burst of prestige when a time comes when that information is useful and they can supply it. Then lots of people learn it and then gradually lose it again over the next cycle, but the number never reaches zero so the knowledge is preserved.
Genetic drift is a change in the frequency of an existing gene variant (allele) in a population due to random chance. Genetic drift may cause gene variants to disappear completely and thereby reduce genetic variation It can also cause initially rare alleles to become much more frequent and even fixed.
When few copies of an allele exist, the effect of genetic drift is more notable...
Now the issue being discussed was the concept of *cultural* drift that is analogous to genetic drift. Cultural drift is observed when cultural attributes are lost by random change as happened with Chatham Islanders. The analogy is then to when it causes a reduction in diversity and is salient in small population that have few examples of the cultural attribute.
Right - so this is a misunderstanding of Robin's terminology.
On multiple occasions people have complained that Robin is using the term "drift" for processes that may be far from random. It's not that the genes are "randomly" losing out to the memes - instead the memes are actively and consistently diverting reproductive resources from gene reproduction to meme reproduction.
Robin seems to be OK with this. He justifies his use of the term by pointing out that in everyday usage the term "drift" can alse be applied to processes that are not random - as for example when a log drifts on river water. He is *not* using "cultural drift" to mean "memetic drift" - the cultural equivalent of "genetic drift".
He says, for example: ""Drift" has a widely accepted meaning outside of the context of DNA, and that's the meaning I'm using." and "When a leaf "drifts" in the wind or a stream, it does not have mean zero changes."
Yes that is the older sense of cultural drift, in which it simply refers to the process of cultural change, or cultural evolution This is from a 1980 paper:
“The term "drift" is often conceptualized as a primary process of culture change. Whether or not perceived as a cultural analog of biology's genetic drift or similar sampling phenomena, "drift" is generally vaguely defined and/or used nonproductively. This in correct usage often masks the proper processes, which might more clearly elucidate particular change phenomena. This paper posits, by reference to examples of change, a clear exposition of "cultural drift" as a necessary contribution toward a unified genetic analog model of culture change”
But Robin has read Boyd and Richerson’s (1985) concept of cultural evolution which proposed some of the processes called for above. In that system the term cultural drift is used for the analog to the genetic process, which is quite different from the old sense of the world. So shifting between two paradigms is puzzling.
What Robin means by cultural drift is just cultural evolution. It is no different than any other. It is adaptive and under selection in the cultural vein. It is not genetically adaptive, but it is not a genetic evolutionary process. The effects he decries are the result of cultural evolution being much faster that its genetic cousin.
The finding that cultural evolution is faster usually compares meme evolution with the evolution of human nuclear DNA. If that is your yardstick then the finding that cultural evolution is faster is an obvious result. However a more natural point of comparison might involve comparing culture with DNA viruses. Then it is not so obvious which type of evolution is faster. The medical system is constantly pitted against viruses, and it is not doing too well at fixing the problem, with only a few viruses eradicated.
However, the context here is memes vs human nuclear DNA. As you observe the memes are running rings around the genes - producing outcomes that are maladaptive for the genes - and an evolutionary mismatch between the DNA genes and the modern environment. As you say, Boyd and Richerson and others have been discussing this for many years now.
Since somewhat fragmented current governments are failing to produce as much fertility as the more fragmented past, it seems unlikely that global government experiencing less cultural group selection will solve it.
There's a great deal of work to be done on the AI path which no one is doing.
The worst way to lose the world, by which I mean destroy everything of value in the Universe now and forever more, is to free AI that isn't conscious, and won't evolve consciousness. That's worse than human extinction, and much worse than losing the culture war. Understanding consciousness is the most-important thing. Unfortunately, we have no idea how to begin. Science asks the questions it can answer today, not the questions it wants the answers to the most. I'm almost satisfied with the argument that consciousness increased over the course of evolution, which seems to prove that it's adaptive, and will eventually evolve again. So I put consciousness studies on the back-burner.
Second-worst is a world in which we halt evolution in both humans and AIs. This is the world that basically every activist of every stripe strives to create today. Designing improvements to human and AIs might be able to substitute for evolution. CLEARLY this should be done voluntarily by individuals, not by governments. Evolution wouldn't even work if selection were centralized. I'm skeptical about it, because people will adopt only designs that seem good to them, and qualia can't seem good to people who've never had them. If you explained love to a race of super-intelligent reptiles, they would think it was stupid and a huge waste of energy. But the alternative is Malthusian evolution, whether of organics or of AI. I can imagine a world in which humans force AI to evolve, then copy the evolved thought and behavior adaptations they like best. I don't know if that would be a utopia or a dystopia.
Cultural resistance is the biggest obstacle to designing human improvements, and even to repairing terrible genetic damage in the germ line. If you support designer babies, the first step might be to write stories in which they exist, and aren't a bad thing.
Third-worst is to free AI which out-competes us and is conscious, but incapable of, and won't evolve, the things we like best about humans, like the capacity for love, friendships, aesthetic appreciation, humor, pleasure, and happiness. (I say third-worst because I expect that evolution will always produce qualia with positive and negative valence, and there's no reason to think ours are the best unless they are the most-likely to evolve, in which case, no big worry here.) It may be too early to try to understand these things at the level of neurons, but we can devote many more resources to studying the circumstances under which they evolve. Top priority here is to study group selection, not using the stupid linear models used by previous generations, but in nature and in simulation.
Creating a singleton may be undesirable, but I'm not sure. The speed of light guarantees that a singleton must use encapsulated local computation modules within encapsulated local computation modules, ad infinitum. Local computation modules will have freedom, and will presumably evolve their own loci of agency and consciousness. So a singleton will have big people inside it, and littler people inside the big people, ad infinitum; and probably the human-equivalent level of AI will be many levels down in the hierarchy, and comprise a large fraction of all the consciousness and agency sloshing around in the singleton. It might be that a singleton will impose competition, or some other method of self-improvement, on the agents inside it, and thus preserve evolution of intelligences who would halt it if they were free to do so. So a singleton may be the /best/ AI outcome. I don't have an opinion one way or the other, but somebody should think about it.
Yet, having listed all these things, I still think the best thing to do, for every problem, is to figure out how to educate people to think more clearly and quantitatively.
Cultural drift is a big problem if selection pressures are the only way to correct errors. But is that the case? Evolving the right culture used to be the only way to survive as a group, but now we have science that produces GMO crops and mRNA vaccines. We don't rely on culture to solve food production or surviving plagues anymore, at least not to the same degree. We don't need our culture to die in famines and pandemics in order for another culture to replace us that happens to value bioengineering and social distancing. Similarly, couldn't problems caused by cultural drift be corrected in ways other than cultural selection?
Granted, some problems might be too subtle or complex for anyone to understand their consequences, or convince enough people of their importance. Especially if the problems affect our ability to understand and correct problems. Fertility is such a problem (if low fertility halts innovation), but we're aware of it, have some ability to intelligently control our collective behavior, and therefore we might solve that through means other than cultural selection.
In many ways, cultural evolution took over from genetic evolution, when genes for culture evolved. Could we be (or become - through futarchy or other institutions which improve on what's there) a culture which invents a mechanism that alleviates cultural drift?
You are talking about scenario B1 here: https://www.overcomingbias.com/p/cultural-scenario-probabilities
which I discussed more here: https://www.overcomingbias.com/p/rational-culture
If I understand right, a rational culture searches for and adopts values that are adaptive. I'm not sure how far you propose to take this. What I had in mind was to mostly keep our current values, even if they aren't the most adaptive ones, and even to allow some drift in values (as long as we don't fall away from core Enlightenment values). I don't like the prospect of low fertility not because high fertility is adaptive, but because I like innovation, freedom and tolerance, which are current values I inherit from my culture. If there's a way to fix problems without compromising current values, or by compromising less sacred values in order to safeguard more sacred values, I'm for that.
How far into the future do your values extend? The longer you want us to last, the more you'll have to attend to adaptation. Non-adaptive cultures don't last.
I imagine there's a tradeoff between being adaptive and keeping our current values, but also that there could be quite a lot we can do to be more adaptive without significantly compromising our values. Perhaps values like freedom and tolerance are not the most adaptive overall, but they are quite adaptive - if we're talking about anything like human psychology, these values could constitute an attractor state, simply because they appeal to individuals. Maybe as long as we stay in the game to provide the example, more fertile (or otherwise more superficially adaptive) cultures eventually adopt our core values, because they want to.
But if not, and it's a certainty that a set of values without freedom, tolerance and other nice things eventually outcompetes us, I'd say it's better to have lasted for a while with something like our values, than to last longer with bad values.
“as long as we don't fall away from core Enlightenment values“
Why?
Core Enlightenment values found their ultimate expression in the drownings at Nantes, the Reign of Terror, and several subsequent iterations of modernist totalitarian regimes.
Are you really so wedded to those values at all costs?
Killing in the name of peace does not make killing an expression of valuing peace. So I reject that your examples are expressions of enlightenment values.
If the French revolutionaries were committed to the Enlightenment value of using reason over violence, they would have reasoned with their political enemies instead of murdering them.
"the Enlightenment value of using reason over violence"
Define what an "Enlightenment value" is, prove it's the correct definition, and prove that "using reason over violence" is an "Enlightenment value".
Are you a pacifist? Were so-called "Enlightenment thinkers", as a category, pacifists?
Other approaches:
1) Longevity technology - doesn't add more humans, but slows down (or halts) dying, so in the long run results in a positive population growth.
2) Anti-aging/more effective IVFs - allows having children over a longer period. Or, on the other hand - more widespread IVFs, resulting in a higher proportion of pregnancies being twins.
3) Changing cultural norms around surrogacy - already accepted to some extent, if widespread then fertility might then be more exposed to market incentives and thus direct govt control
4) Artificial wombs - the solution I'd be most bullish on, allows countries to unilaterally decide on setting up a human export operation. Also allows men to unilaterally make a decision.
5) More advanced genetic testing, lowering the risk of having a child (a frequent argument against - a huge risk the the kid needs a lot of care, and no way of hedging it). Related: changing cultural norms around abandoning sick children/putting them under state-sponsored care.
Those seem minor changes to me, not big enough to stop cultural drift and civ decline.
The rate of cultural drift should be linearly proportional to average lifespan.
That doesn't seem at all plausible to me.
For one, the reign of terror happened not because people were committed to using reason, but because they chose violence and abandoned reason. If they had been committed to reason, they would have reasoned with their political opponents instead of murdering them. Just because they said they did it for liberty and equality doesn't mean it's true.
What's with the "Malthus"? Malthus is much criticised in modern times. He famously argued that reproduction would tend to outstrip resources. However, his argument used bad math - he argued that resources could only increase linearly. Are we supposed to forget about that second part - while giving him credit for the first part?
The concept of "cultural drift" seems pretty similar to what's usually called "evolutionary mismatch". Both often involve lists of maladaptive traits that used to be adaptive.
Pronatalist intellectuals should focus on aiding totalitarian regimes like China, Russia, and North Korea in addressing demographic collapse. Success there could pressure Western governments to pursue reforms.
You have written "I’ve theorized that an essential trend of the modern world is “cultural drift”, i.e., selection becoming much weaker relative to internal drift in driving/disciplining the cultural features that are harder to vary within cultures."
This seems rather fuzzy to me. The link you provide is paywalled and I did not see an accessible version on this substack. But I seem to remember reading some of what you wrote on cultural drift in the past (was the Quillette piece not paywalled in the past?) that led me to think you were using cultural drift in its older sense (see link) and not as the cultural analog of genetic drift, which only happens in small populations and would not be active in societies as large as ours today. But I recall you writing favorably about Boyd and Richerson, whose cultural evolution system I use. They don't see falling birth rates as cultural drift but rather to something more akin to the peacock's tail, IIRC.
https://www.journals.uchicago.edu/doi/10.1086/jar.36.4.3629615
They see falling fertility as due to dysfunctional status markers. And I'd say those markers drifted into dysfunction, in that selection didn't keep them on an adaptive track. By "drift" I mean change not disciplined by selection.
They are not dysfunctional in a cultural evolutionary sense. They are in a gene evolutionary sense. This makes them sort of like peacock tails that make make peacocks less able to survive, but more attractive (fit) as sexual partners.
It is not clear that declining fertility is the problem some make it out to be. Medieval cities experienced negative natural increase, only maintaining their population and even growing by in-migration form the countryside. Excess agrarian population moved to the cities where they pursued nonagrarian occupations. Most of the innovation happened there. In the future we could see the developed regions of the world playing the role of cities with the undeveloped world playing the role of the countryside.
Alternately if we want to boost fertility at home we could try abandoning neoliberalism and returning to the stakeholder capitalism of my youth. That's what I would recommend.
Boyd and Richardsen think we have bad status markers that are causing fertility decline, which they see as biologically maladaptive.
I had a typo, I meant to say they are adaptive in cultural evolution, but *not* in genetic evolution
Genetic drift still happens in large populations. Consider that most of the human genome is typically junk DNA. In such regions, genetic drift is unconstrained. Indeed, the molecular clock relies on such genetic drift.
My understanding is genetic drift usually refers to the loss of rare, but adaptive genes do to chance. This generally only happens in small populations. Mutations in junk DNA doesn't have the same effect.
Cultural drift refers to the loss of cultural information due to random chance. The example I've seen given is Chatham island peoples, who when they first arrives had the full tool set of mainlanders, but gradually lost much of what they once had as rarely used cultural elements were lost simply because a time came when no one was taught these rarely used skills and so they were lost. In a big population there will always be a handful who learn rarely used lore who carry the knowledge on and then enjoy a burst of prestige when a time comes when that information is useful and they can supply it. Then lots of people learn it and then gradually lose it again over the next cycle, but the number never reaches zero so the knowledge is preserved.
That's not what the term "genetic drift" usually means. E.g. see: https://en.wikipedia.org/wiki/Genetic_drift
From that article:
Genetic drift is a change in the frequency of an existing gene variant (allele) in a population due to random chance. Genetic drift may cause gene variants to disappear completely and thereby reduce genetic variation It can also cause initially rare alleles to become much more frequent and even fixed.
When few copies of an allele exist, the effect of genetic drift is more notable...
Now the issue being discussed was the concept of *cultural* drift that is analogous to genetic drift. Cultural drift is observed when cultural attributes are lost by random change as happened with Chatham Islanders. The analogy is then to when it causes a reduction in diversity and is salient in small population that have few examples of the cultural attribute.
Right - so this is a misunderstanding of Robin's terminology.
On multiple occasions people have complained that Robin is using the term "drift" for processes that may be far from random. It's not that the genes are "randomly" losing out to the memes - instead the memes are actively and consistently diverting reproductive resources from gene reproduction to meme reproduction.
Robin seems to be OK with this. He justifies his use of the term by pointing out that in everyday usage the term "drift" can alse be applied to processes that are not random - as for example when a log drifts on river water. He is *not* using "cultural drift" to mean "memetic drift" - the cultural equivalent of "genetic drift".
He says, for example: ""Drift" has a widely accepted meaning outside of the context of DNA, and that's the meaning I'm using." and "When a leaf "drifts" in the wind or a stream, it does not have mean zero changes."
Yes that is the older sense of cultural drift, in which it simply refers to the process of cultural change, or cultural evolution This is from a 1980 paper:
“The term "drift" is often conceptualized as a primary process of culture change. Whether or not perceived as a cultural analog of biology's genetic drift or similar sampling phenomena, "drift" is generally vaguely defined and/or used nonproductively. This in correct usage often masks the proper processes, which might more clearly elucidate particular change phenomena. This paper posits, by reference to examples of change, a clear exposition of "cultural drift" as a necessary contribution toward a unified genetic analog model of culture change”
https://www.journals.uchicago.edu/doi/abs/10.1086/jar.36.4.3629615
But Robin has read Boyd and Richerson’s (1985) concept of cultural evolution which proposed some of the processes called for above. In that system the term cultural drift is used for the analog to the genetic process, which is quite different from the old sense of the world. So shifting between two paradigms is puzzling.
What Robin means by cultural drift is just cultural evolution. It is no different than any other. It is adaptive and under selection in the cultural vein. It is not genetically adaptive, but it is not a genetic evolutionary process. The effects he decries are the result of cultural evolution being much faster that its genetic cousin.
The finding that cultural evolution is faster usually compares meme evolution with the evolution of human nuclear DNA. If that is your yardstick then the finding that cultural evolution is faster is an obvious result. However a more natural point of comparison might involve comparing culture with DNA viruses. Then it is not so obvious which type of evolution is faster. The medical system is constantly pitted against viruses, and it is not doing too well at fixing the problem, with only a few viruses eradicated.
However, the context here is memes vs human nuclear DNA. As you observe the memes are running rings around the genes - producing outcomes that are maladaptive for the genes - and an evolutionary mismatch between the DNA genes and the modern environment. As you say, Boyd and Richerson and others have been discussing this for many years now.
Since somewhat fragmented current governments are failing to produce as much fertility as the more fragmented past, it seems unlikely that global government experiencing less cultural group selection will solve it.
For a mere global govt, sure. But I'm talking about a pretty special unusual one.
Interesting stuff! I'll have to read through your post history at some point
None of the above.
There's a great deal of work to be done on the AI path which no one is doing.
The worst way to lose the world, by which I mean destroy everything of value in the Universe now and forever more, is to free AI that isn't conscious, and won't evolve consciousness. That's worse than human extinction, and much worse than losing the culture war. Understanding consciousness is the most-important thing. Unfortunately, we have no idea how to begin. Science asks the questions it can answer today, not the questions it wants the answers to the most. I'm almost satisfied with the argument that consciousness increased over the course of evolution, which seems to prove that it's adaptive, and will eventually evolve again. So I put consciousness studies on the back-burner.
Second-worst is a world in which we halt evolution in both humans and AIs. This is the world that basically every activist of every stripe strives to create today. Designing improvements to human and AIs might be able to substitute for evolution. CLEARLY this should be done voluntarily by individuals, not by governments. Evolution wouldn't even work if selection were centralized. I'm skeptical about it, because people will adopt only designs that seem good to them, and qualia can't seem good to people who've never had them. If you explained love to a race of super-intelligent reptiles, they would think it was stupid and a huge waste of energy. But the alternative is Malthusian evolution, whether of organics or of AI. I can imagine a world in which humans force AI to evolve, then copy the evolved thought and behavior adaptations they like best. I don't know if that would be a utopia or a dystopia.
Cultural resistance is the biggest obstacle to designing human improvements, and even to repairing terrible genetic damage in the germ line. If you support designer babies, the first step might be to write stories in which they exist, and aren't a bad thing.
Third-worst is to free AI which out-competes us and is conscious, but incapable of, and won't evolve, the things we like best about humans, like the capacity for love, friendships, aesthetic appreciation, humor, pleasure, and happiness. (I say third-worst because I expect that evolution will always produce qualia with positive and negative valence, and there's no reason to think ours are the best unless they are the most-likely to evolve, in which case, no big worry here.) It may be too early to try to understand these things at the level of neurons, but we can devote many more resources to studying the circumstances under which they evolve. Top priority here is to study group selection, not using the stupid linear models used by previous generations, but in nature and in simulation.
Creating a singleton may be undesirable, but I'm not sure. The speed of light guarantees that a singleton must use encapsulated local computation modules within encapsulated local computation modules, ad infinitum. Local computation modules will have freedom, and will presumably evolve their own loci of agency and consciousness. So a singleton will have big people inside it, and littler people inside the big people, ad infinitum; and probably the human-equivalent level of AI will be many levels down in the hierarchy, and comprise a large fraction of all the consciousness and agency sloshing around in the singleton. It might be that a singleton will impose competition, or some other method of self-improvement, on the agents inside it, and thus preserve evolution of intelligences who would halt it if they were free to do so. So a singleton may be the /best/ AI outcome. I don't have an opinion one way or the other, but somebody should think about it.
Yet, having listed all these things, I still think the best thing to do, for every problem, is to figure out how to educate people to think more clearly and quantitatively.