“There’s an invisible hand, so to speak, that will push parents to select for children with nearly identical traits, resulting in a radical homogenization of society where straight white males will dominate even more than they do right now.”
So, a field of uniform, tall, genetically-identical corn, as far as the eye can see.
You know, that kind of monoculture does especially bad when corn blight hits.
Most of the fear here comes from treating AI as an epistemic shock rather than a coordination shock.
When intelligence becomes abundant, the scarce resource becomes orientation.
Students offload thinking, researchers overclaim, startups promise impossible timelines, and parents fantasize about engineered children not because the actors are irrational, but because the system is desynchronized.
Meaning collapses first.
Institutions collapse second.
Not because of AI’s capability, but because of AI’s speed.
“How many of you know of someone who used AI to assist in writing at least one assignment?”
Interesting anecdote about the 3/4 of students raising their hands, but it seems like it all hinges on how the phrase "assist in writing" is interpreted. I suspect you meant "assist with composition," as in students using AI-generated words, sentences and paragraphs. That's clearly cheating (and stupid) and so 3/4 would be a pretty depressing finding.
On the other hand, what if some students interpreted "assist" to mean merely assistance with fact-checking and research? That would be more acceptable, I'd think, so long as they heeded your warning about the "reliable unreliability" of AI and checked the LLM's answers, investigated its citations, etc. Anyway, just a thought...
My sense is that it was a mix of both -- you're right, I could have been clearer. I do think that some students incorporated AI-generated sentences into their assignments, whereas others (or those same students at other times) simply used AI to get info, etc. Thanks for pointing this out!!
The fallacy behind “designer babies” is the assumption that allowing parents to rationally choose their children’s traits somehow frees humanity from natural selection. It doesn’t. Human preferences themselves are products of natural and sexual selection, shaped over long periods by the coevolution of traits and the preferences for those traits.
Designer-baby selection doesn’t eliminate this process; it simply moves selection from mate choice to trait choice. The evolutionary game continues, just with a new mechanism. And once you recognise that preferences have been shaped by evolution just as much as bodies have, the idea of “rational” selection becomes much less straightforward.
A key point often missed is that genes and preferences can coevolve. Throughout human evolution, certain traits and the preferences for those traits have reinforced one another in classic Fisherian feedback loops. For example, uniquely human features such as continuous scalp-hair growth or extreme reduction of body hair in females appear to have persisted not because they improve survival, but because mate preferences favoured them — and those preferences themselves show heritable variation. In such cases, genes shaping a visible trait and genes shaping the preference for that trait spread together, independent of any environmental advantage.
This becomes especially relevant in a designer-baby regime. From a gene’s-eye view, what matters is simply being copied. Genes that produce traits people prefer — and genes that influence those preferences — flourish under any system where humans choose which genomes move forward. And that means the outcomes may not resemble what techno-eugenicists imagine.
Consider a hypothetical gene complex that codes simultaneously for
(a) bright blue skin,
(b) a strong preference for children with bright blue skin, and
(c) low IQ.
Under a designer-baby system, such a gene could spread rapidly simply because it shapes both the trait and the preference for the trait. The low-IQ component wouldn’t stop it. If the preference loop dominates the selection process, the gene prospers. This example illustrates the gap between what eugenicists think they’re selecting for and what evolution actually selects, given that preferences are themselves part of the system.
In short: designer babies don’t replace natural selection. They become part of it. And once preference–trait feedback loops start driving reproductive decisions, the evolutionary trajectory becomes unpredictable — and quite possibly hostile to the utopian visions of contemporary eugenics enthusiasts.
“There’s an invisible hand, so to speak, that will push parents to select for children with nearly identical traits, resulting in a radical homogenization of society where straight white males will dominate even more than they do right now.”
So, a field of uniform, tall, genetically-identical corn, as far as the eye can see.
You know, that kind of monoculture does especially bad when corn blight hits.
Most of the fear here comes from treating AI as an epistemic shock rather than a coordination shock.
When intelligence becomes abundant, the scarce resource becomes orientation.
Students offload thinking, researchers overclaim, startups promise impossible timelines, and parents fantasize about engineered children not because the actors are irrational, but because the system is desynchronized.
Meaning collapses first.
Institutions collapse second.
Not because of AI’s capability, but because of AI’s speed.
Authorship prioritization ?
Outsourcing thinking isn’t required.
“How many of you know of someone who used AI to assist in writing at least one assignment?”
Interesting anecdote about the 3/4 of students raising their hands, but it seems like it all hinges on how the phrase "assist in writing" is interpreted. I suspect you meant "assist with composition," as in students using AI-generated words, sentences and paragraphs. That's clearly cheating (and stupid) and so 3/4 would be a pretty depressing finding.
On the other hand, what if some students interpreted "assist" to mean merely assistance with fact-checking and research? That would be more acceptable, I'd think, so long as they heeded your warning about the "reliable unreliability" of AI and checked the LLM's answers, investigated its citations, etc. Anyway, just a thought...
My sense is that it was a mix of both -- you're right, I could have been clearer. I do think that some students incorporated AI-generated sentences into their assignments, whereas others (or those same students at other times) simply used AI to get info, etc. Thanks for pointing this out!!
Humans will colonize Mars...excellent. Go!!!!!!!!!!!!!!!
The fallacy behind “designer babies” is the assumption that allowing parents to rationally choose their children’s traits somehow frees humanity from natural selection. It doesn’t. Human preferences themselves are products of natural and sexual selection, shaped over long periods by the coevolution of traits and the preferences for those traits.
Designer-baby selection doesn’t eliminate this process; it simply moves selection from mate choice to trait choice. The evolutionary game continues, just with a new mechanism. And once you recognise that preferences have been shaped by evolution just as much as bodies have, the idea of “rational” selection becomes much less straightforward.
A key point often missed is that genes and preferences can coevolve. Throughout human evolution, certain traits and the preferences for those traits have reinforced one another in classic Fisherian feedback loops. For example, uniquely human features such as continuous scalp-hair growth or extreme reduction of body hair in females appear to have persisted not because they improve survival, but because mate preferences favoured them — and those preferences themselves show heritable variation. In such cases, genes shaping a visible trait and genes shaping the preference for that trait spread together, independent of any environmental advantage.
This becomes especially relevant in a designer-baby regime. From a gene’s-eye view, what matters is simply being copied. Genes that produce traits people prefer — and genes that influence those preferences — flourish under any system where humans choose which genomes move forward. And that means the outcomes may not resemble what techno-eugenicists imagine.
Consider a hypothetical gene complex that codes simultaneously for
(a) bright blue skin,
(b) a strong preference for children with bright blue skin, and
(c) low IQ.
Under a designer-baby system, such a gene could spread rapidly simply because it shapes both the trait and the preference for the trait. The low-IQ component wouldn’t stop it. If the preference loop dominates the selection process, the gene prospers. This example illustrates the gap between what eugenicists think they’re selecting for and what evolution actually selects, given that preferences are themselves part of the system.
In short: designer babies don’t replace natural selection. They become part of it. And once preference–trait feedback loops start driving reproductive decisions, the evolutionary trajectory becomes unpredictable — and quite possibly hostile to the utopian visions of contemporary eugenics enthusiasts.
Is Grok 5 just 3 weeks away?
Good luck in Europe Émile! As long as you continue the Zoom calls on the other side of the pond, it’ll be like you never even left.
Um… solid work I’m restacking
Gl on your new home!