Such a great piece! Thank you for highlighting the human aspect in this crazy AI era. Chinese Gen Z is definitely having a hard time right now, especially college kids and even PhD graduates who did everything they were told, only to wake up in this scary job market.
It’s true that things have always been super competitive in China, but at least people with advanced degrees used to find better jobs much more easily. A large part of the current struggle is due to the post-COVID economic downturn, as well as the continuously rising number of highly educated young people that you pointed out in the article. AI is just hitting everything even harder.
Like many people in China, I don’t think we can avoid the AI era. Unlike some in the West who are actively trying to push back against AI, we feel like we have to embrace it, otherwise, we’ll be left behind. So, the real question is whether we can adapt our system to suit this drastic change and actually benefit people. That is the question of the century.
The "full-time children" detail isn't that surprising, with Gen Z riding on the boom generation's savings in the most literal possible way and the government having every incentive to not call it what it is. Reminds me of the laying flat 躺平 trend in China.
I'm glad you closed with the Carl Benedikt Frey quote too. "The short run can be a lifetime" makes the optimist case feel incomplete every time.
Sharper parallel than it looks. China named 躺平 directly and let it become political; "full-time children" is the euphemism that keeps the same dynamic readable only as a family choice. Naming difference may matter more than the behavior itself.
Appreciate the expansion pack! Exciting to see your path and your growth as a writer.
I´m a college student in Chile, a country that also underwent mass college enrollment in the 2000s. Not sure what AI means for young people in middle income countries.
As important income is, I also worry about the loss of meaning jobs provide. They provide social legibility, a way to feel useful as an adult.
Hannah Arendt wrote about the active and contemplative life. Maybe we need a new category?
+
What will the kind of people who used to go to software engineering do? What did these types of people do before white collar jobs? Academia?
An absolutely fascinating glimpse @Jasmine. It’s interesting how the different social constructs in China and say the US (or Western Europe) create such diverse expectations of the impending AI revolution. I have to say that at 63, my hope for my 5 year old granddaughter leans optimistic, but I can assure young people who are 19 now, that when I was 19 in 1981 the world economy was in shocking state and I could not find employment or afford higher education. I hasten to add that the last 44 years have been periods of adaptation against many technical innovation cycles.
Never stop learning, adaptations, and try to stay relevant. Even now at my age, I use the prosumer AI tools and try to share my take on what they mean, with a healthy dose of skeptical optimism. :-)
Great piece as always ! It was also fascinating to hear some of your thoughts on the China trip. My own impression from my last China trip in 2024 was that joblessness among college graduates was already a large topic though the AI component was missing. Many people I know at that age had considerably lower expectations or pivoted way (one from physics to accounting). The anxiety and fear around jobs permeated many conversations.
I agree with your point on tools and some of the other points, but am quite sceptical of the UBI idea. My impression is that some amount of distortion of AI companies objectives (through a form of tax) might be necessary. When it comes to the distribution it matters a lot how permanently the economy is changed. If retraining is for example not an option for many anymore it would be more like a permanent subsidy, which needs to be financed differently. Anyway despite your critical view of econ, I think you'd make an excellent economist 😅
yeah I'm a bit skeptical of UBI too, especially if some jobs are automated sooner than others (also I think people need more structure - so at minimum it's necessary but not sufficient)
haha I'm not anti-economist, I read so much econ stuff for this piece! but as a sociology person my job is also to point out what they might miss
I think the U.S. income tax system will need to evolve if fewer human workers are needed in future - some kind of token tax? Ideally, then a rapid population decrease due to low birth rates in developed countries won‘t be as socially disruptive as feared. One problem I fear is that worldwide technology advancement will be very uneven; not every country will progress as rapidly as China. Much of Africa and even India don’t yet have reliable infrastructure, and many regions lack political stability and consistent rule of law. AI may thus end up increasing the disparity between the developed and the developing world, and immigrants from left-behind regions will be even less welcome than now if their labor is not wanted at any price.
> But Klein’s example also reveals why I am less rosy than he is.
A bigger problem with Klein's examples is that they *don't* demonstrate a "human premium" at all!
A 'human premium' is when wealth-owners put an intrinsic value on a human doing a service, even if the human does it worse or more expensive than the alternative automation of the *same service*. This then defeats any argument about efficiency etc, because there is perpetual employment for humans just for being humans no matter how objectively inferior. So to show a human premium, you need to show many examples of cases where humans do the job worse or more expensive than the AI alternative but there is durable demand for those humans anyway. This means, most obviously, that the AI alternative must not just exist but be equally good or superior. You cannot express a preference for human versions of services if those are the only versions... (Did Ford's customers have an "intrinsic black preference"? Subsequent history suggests they did not.)
And none of Klein's examples are that. Claude can't prescribe you drugs for your allergy or give you an official diagnosis. So the doctor isn't a 'human premium'. It is controversial, to say the least, that Claude is a superhuman therapist better than Klein's therapist. So the therapist isn't a human premium. And there are no superhuman video editors. So those video editors aren't a human premium either. (And so on for the examples like Starbucks: a few ounces of boiling liquid is *not* the total package Starbucks provides, therefore it is irrelevant if your home espresso machine nominally does that small part equally well and it doesn't show a human premium.)
This is also why Imas's point is wrong. The history of richer economies buying more human services cannot show that people prefer human services over AI services - because the relevant AI have not existed, and generally still do not.
Good points — Alex similarly clarified to me what he meant by “human premium” / “relational good”, and my reaction to that was like “man I do not think there are a lot of goods/services that fall into that category.” Arguably the artisan goods might hold a human premium under your definition, if one is willing to pay a big markup for the little label that says “Made with <3 by women ceramicists in Oaxaca” or whatever over the good Chinese dupe. But again, I wonder how big the market for that really is, assuming that the human version is also less convenient to obtain (maybe Alex would say that inconvenience = more valuable, like Birkin bags and popular restaurants, but AIs can create artificial scarcity too).
I'm not sure there are any, because I notice that every time we get genuinely superhuman performance, the human employment seems to gradually shrink over time, and we have no examples of long-run human premia aside from cases which appear to be essentially just celebrity fandoms with the nominal job as a thin excuse. (Magnus Carlsen's fans are just as happy to see him do fantasy football as they are chess. Does this prove that there is a human premium for chess, or a human premium large enough to matter macroeconomically for all of humanity indefinitely in a perpetual AI future for centuries or millennia to come?)
Anyway, another big problem with the 'human premium' idea is that it assumes that human premia are things which *can* survive in such a future economy, as if they existed in a vacuum unaffected by their context. You give a good example: maybe there is a human premium for the intangible of being made by hand. But... as you point out, it's actually quite hard to buy those things, rather than Etsy dropships of things lovingly manufactured by robots in Shenzhen, because the good Chinese dupe may well be of superior quality objectively and its very cheapness incentivizes the market to work around your 'human premium' by giving you the illusion of it. That little label may cost $1000 over the highly optimized lights out factory making it for $1, which means that you now have $999 of incentive to figure out how to spam, advertise, pay off influencers, hack, bribe, set up deepfakes of the creators livestreaming 'their' wares, or whatever it takes. (Pigbutchering scams, for example, sometimes provide beautiful young women on video calls to victims to prove that it's not a scam, by simply hiring one real woman to do 5-minute calls all day long, which means that you could do weekly calls for <1000 victims per woman. They can afford many things like that because their supposed product, cryptocurrency/option trading, is so fake and is so high margin.) The way that 'fair trade' or 'carbon credits' or 'bean to bar' or 'farm to table' are dogged with fraud (sometimes approaching 100%), because it's a lot cheaper to create the illusion than the reality, and most consumers turn out to not have *that* much of a premium for the reality. And the reality of this lack of reality slowly corrodes away enthusiasm for it, eliminating that intangible's premium. A century later...
What examples do we have of genuinely superhuman performance aside from Chess? I'm not sure we have enough to generalize from. There are a variety of examples of historical automation with different dynamics (bank tellers, accountants upon the introduction of spreadsheet software).
I think it's hard to say in part because there is no human premium, so as soon as we attain truly superhuman performance on a whole task, not a single subpart, it is quickly defined away and becomes nothing but an amusing footnote in history texts about absurdly inefficient and wasteful ways to do something, if that.
Have you ever paid a 'knocker-upper' lately to wake you up? How many people have you hired to operate an old HP desk calculator for you of late? "No, of course not, Gwern, that would be ludicrous, I have an alarm on my smartphone and I have no use for a 'computer' to do arithmetic for me when I have digital computers. Why would I want to hire a human for any of that?" Yeah, exactly. You don't even realize you don't have a 'human premium' for countless tasks you take for granted that a machine always does now, even though they *didn't* always do them. You don't have a human premium for harvesting corn in Iowa by hand, either, or pretty much anything done by a machine.
Notice that in the only examples you gave, machines *don't* clearly have superhuman performance, because bank tellers and accountants can do a lot more than just a few fragmented tasks that ATMs and spreadsheets can do.
(I personally would not because I don’t think it would work. See: Ukraine. Also, as a Dune fan, I must point out that the Butlerian Jihad is widely misunderstood, in large part due to the wretched sequels/prequels, and we face a scenario closer to the end of _Chapterhouse: Dune_, rather than the Butlerian Jihad.)
I meant more in the sense that if your arguments are true then the only ethical option would be a crusade against computers, not that this is what you were actually suggesting.
I also disagree with that. I don’t think it’s moral to do things which don’t work and make the world a worse place; I think it’s moral to do things which make the world a better place.
Yea I found it weird that both Ezra and Imas didn't really address this fairly straightforward point:
"Just because something is scarce and high-status doesn’t mean that everyone will be able to afford it; in fact, many people won’t."
We're assuming that most people will be able to afford all these highly priced relational goods on the basis of what...that more people will be in high-income relational jobs? Or because commodity prices will drop?
Reminds me of de-industrialization -- your community crumbled when the factory closed, but it's gonna be ok because that was offset by cheap tvs. There are going to be a lot of losers. And the richest 10% already make up over half of consumer spending. So this sounds right:
"So we may end up in a world where most people are handmaidens to the super-rich, while most normal people settle for cheap digital alternatives."
Thank you for helping me understand AI and the impact it will have on the future of work. The pragmatism China has demonstrated and the realisation that it is a tool that will impact society in massive ways.... I liked the argument that AI has cost determinants in China and can see how the argument will develop in the future. The argument reminds me of the birth of digital and the revolt against mechanisation in the nineteenth century cotton mills in the UK... there are so many parallels throughout history... It is a wonderful article... thank you....
1:1 teacher student ratio would be great for when student population drops with population to match the people who want to be teachers and are reasonably good at teaching, but education is a whole other can of worms without AI in the mix.
The full-time children feels like a compromise on traditional family dynamics--in the past, the eldest son or daughter's family might have been expected to care for the parents (depends on regional traditions), but sometimes a kid who doesn't marry might perform this role. In a family with many children who survive to adulthood, you might have some complex family inheritance wrangling mixed in for who gets to provide the care (for good or ill) so they control the parents' wealth. I have some experience with this from some older aunts/uncles who had many children (4-9) in Shanghai, though I also know some happy and drama-free families too.
Regarding jobs generally, there will always be headaches in how people see their wage based on endogenous factors that will just be different across many people. What do they want, what do they need, are there food, restaurant, leisure, clothing, whatever brands that they absolutely must have? What is their cost of living where they live? Maybe they can't move to a cheaper place because they don't have a car or a child or another dependent or partner can't move. Maybe they are stuck somewhere because of the peculiarities of tax or property laws we have on state and local levels in the US. A person happy with a multigenerational/multiple wage family in one house situation might be able to save a lot more than a single person living in an RV park. Remember that Jevon's Paradox came out of 1865... How people lived (larger families, people migrating to industrial areas may have had hometown migrants to help each other) and spent money (less leisure/goods, many things also more expensive) on was a lot different than where we are today (though it might feel familiar for many Chinese and others in developing countries).
Great piece! The demand-dampening point is what the Jevons optimists keep skating past. Cheaper production only expands demand if the wealth circulates broadly enough to generate it. If it concentrates, you hit a ceiling. The top 1% can't consume their way to replacing middle-class spending. China already shows this: massive productive capacity, weak domestic consumption, high savings rate. It's not a theoretical risk.
The doomsday prophecy of the "permanent underclass" is not something a vague "societal choice" can reverse. Silicon Valley is not "bracing" (what an egregious, disingenuous headline. Assuming NYT did it) but gleeful (its CEOs and owners def seem to be) of their own 'can't be helped' belt-tightening, unchecked power, speculation and prospects.
Long way of saying, I wish your piece had started with reversing the inevitability of these death knell "permanent underclass" proclamations (know you tried to later in the piece). Clearly, the smart researchers and scientists you spoke to over months, had practical ideas besides uncontrolled civil upheaval and the bad guys winning.
I pretty much agree with Jasmine. The irony for the AI originators is that the riches they make will require state regulation as in China to re-distribute among the out of work poor that AI will create. If not then no barricade will keep out a bored and hungry mob. If I had to make a prediction it would be that the next US government will likely be run by a Socialist President voted in by the jobless.
super interesting as usual. I didn't realize you are a free-lancer. I hope you can make a livelihood and keep doing this. I'm really grateful for your work.
Such a great piece! Thank you for highlighting the human aspect in this crazy AI era. Chinese Gen Z is definitely having a hard time right now, especially college kids and even PhD graduates who did everything they were told, only to wake up in this scary job market.
It’s true that things have always been super competitive in China, but at least people with advanced degrees used to find better jobs much more easily. A large part of the current struggle is due to the post-COVID economic downturn, as well as the continuously rising number of highly educated young people that you pointed out in the article. AI is just hitting everything even harder.
Like many people in China, I don’t think we can avoid the AI era. Unlike some in the West who are actively trying to push back against AI, we feel like we have to embrace it, otherwise, we’ll be left behind. So, the real question is whether we can adapt our system to suit this drastic change and actually benefit people. That is the question of the century.
The "full-time children" detail isn't that surprising, with Gen Z riding on the boom generation's savings in the most literal possible way and the government having every incentive to not call it what it is. Reminds me of the laying flat 躺平 trend in China.
I'm glad you closed with the Carl Benedikt Frey quote too. "The short run can be a lifetime" makes the optimist case feel incomplete every time.
Great piece as always.
Sharper parallel than it looks. China named 躺平 directly and let it become political; "full-time children" is the euphemism that keeps the same dynamic readable only as a family choice. Naming difference may matter more than the behavior itself.
Appreciate the expansion pack! Exciting to see your path and your growth as a writer.
I´m a college student in Chile, a country that also underwent mass college enrollment in the 2000s. Not sure what AI means for young people in middle income countries.
As important income is, I also worry about the loss of meaning jobs provide. They provide social legibility, a way to feel useful as an adult.
Hannah Arendt wrote about the active and contemplative life. Maybe we need a new category?
+
What will the kind of people who used to go to software engineering do? What did these types of people do before white collar jobs? Academia?
An absolutely fascinating glimpse @Jasmine. It’s interesting how the different social constructs in China and say the US (or Western Europe) create such diverse expectations of the impending AI revolution. I have to say that at 63, my hope for my 5 year old granddaughter leans optimistic, but I can assure young people who are 19 now, that when I was 19 in 1981 the world economy was in shocking state and I could not find employment or afford higher education. I hasten to add that the last 44 years have been periods of adaptation against many technical innovation cycles.
Never stop learning, adaptations, and try to stay relevant. Even now at my age, I use the prosumer AI tools and try to share my take on what they mean, with a healthy dose of skeptical optimism. :-)
Great piece as always ! It was also fascinating to hear some of your thoughts on the China trip. My own impression from my last China trip in 2024 was that joblessness among college graduates was already a large topic though the AI component was missing. Many people I know at that age had considerably lower expectations or pivoted way (one from physics to accounting). The anxiety and fear around jobs permeated many conversations.
I agree with your point on tools and some of the other points, but am quite sceptical of the UBI idea. My impression is that some amount of distortion of AI companies objectives (through a form of tax) might be necessary. When it comes to the distribution it matters a lot how permanently the economy is changed. If retraining is for example not an option for many anymore it would be more like a permanent subsidy, which needs to be financed differently. Anyway despite your critical view of econ, I think you'd make an excellent economist 😅
yeah I'm a bit skeptical of UBI too, especially if some jobs are automated sooner than others (also I think people need more structure - so at minimum it's necessary but not sufficient)
haha I'm not anti-economist, I read so much econ stuff for this piece! but as a sociology person my job is also to point out what they might miss
That's fair, econ can do with critique. An uncomfortable economist is a good economist (Keynes probably)
I think the U.S. income tax system will need to evolve if fewer human workers are needed in future - some kind of token tax? Ideally, then a rapid population decrease due to low birth rates in developed countries won‘t be as socially disruptive as feared. One problem I fear is that worldwide technology advancement will be very uneven; not every country will progress as rapidly as China. Much of Africa and even India don’t yet have reliable infrastructure, and many regions lack political stability and consistent rule of law. AI may thus end up increasing the disparity between the developed and the developing world, and immigrants from left-behind regions will be even less welcome than now if their labor is not wanted at any price.
100% re: disparity between countries! the US and China are much better equipped to at least have AI companies they can redistribute wealth from
> But Klein’s example also reveals why I am less rosy than he is.
A bigger problem with Klein's examples is that they *don't* demonstrate a "human premium" at all!
A 'human premium' is when wealth-owners put an intrinsic value on a human doing a service, even if the human does it worse or more expensive than the alternative automation of the *same service*. This then defeats any argument about efficiency etc, because there is perpetual employment for humans just for being humans no matter how objectively inferior. So to show a human premium, you need to show many examples of cases where humans do the job worse or more expensive than the AI alternative but there is durable demand for those humans anyway. This means, most obviously, that the AI alternative must not just exist but be equally good or superior. You cannot express a preference for human versions of services if those are the only versions... (Did Ford's customers have an "intrinsic black preference"? Subsequent history suggests they did not.)
And none of Klein's examples are that. Claude can't prescribe you drugs for your allergy or give you an official diagnosis. So the doctor isn't a 'human premium'. It is controversial, to say the least, that Claude is a superhuman therapist better than Klein's therapist. So the therapist isn't a human premium. And there are no superhuman video editors. So those video editors aren't a human premium either. (And so on for the examples like Starbucks: a few ounces of boiling liquid is *not* the total package Starbucks provides, therefore it is irrelevant if your home espresso machine nominally does that small part equally well and it doesn't show a human premium.)
This is also why Imas's point is wrong. The history of richer economies buying more human services cannot show that people prefer human services over AI services - because the relevant AI have not existed, and generally still do not.
Good points — Alex similarly clarified to me what he meant by “human premium” / “relational good”, and my reaction to that was like “man I do not think there are a lot of goods/services that fall into that category.” Arguably the artisan goods might hold a human premium under your definition, if one is willing to pay a big markup for the little label that says “Made with <3 by women ceramicists in Oaxaca” or whatever over the good Chinese dupe. But again, I wonder how big the market for that really is, assuming that the human version is also less convenient to obtain (maybe Alex would say that inconvenience = more valuable, like Birkin bags and popular restaurants, but AIs can create artificial scarcity too).
I'm not sure there are any, because I notice that every time we get genuinely superhuman performance, the human employment seems to gradually shrink over time, and we have no examples of long-run human premia aside from cases which appear to be essentially just celebrity fandoms with the nominal job as a thin excuse. (Magnus Carlsen's fans are just as happy to see him do fantasy football as they are chess. Does this prove that there is a human premium for chess, or a human premium large enough to matter macroeconomically for all of humanity indefinitely in a perpetual AI future for centuries or millennia to come?)
Anyway, another big problem with the 'human premium' idea is that it assumes that human premia are things which *can* survive in such a future economy, as if they existed in a vacuum unaffected by their context. You give a good example: maybe there is a human premium for the intangible of being made by hand. But... as you point out, it's actually quite hard to buy those things, rather than Etsy dropships of things lovingly manufactured by robots in Shenzhen, because the good Chinese dupe may well be of superior quality objectively and its very cheapness incentivizes the market to work around your 'human premium' by giving you the illusion of it. That little label may cost $1000 over the highly optimized lights out factory making it for $1, which means that you now have $999 of incentive to figure out how to spam, advertise, pay off influencers, hack, bribe, set up deepfakes of the creators livestreaming 'their' wares, or whatever it takes. (Pigbutchering scams, for example, sometimes provide beautiful young women on video calls to victims to prove that it's not a scam, by simply hiring one real woman to do 5-minute calls all day long, which means that you could do weekly calls for <1000 victims per woman. They can afford many things like that because their supposed product, cryptocurrency/option trading, is so fake and is so high margin.) The way that 'fair trade' or 'carbon credits' or 'bean to bar' or 'farm to table' are dogged with fraud (sometimes approaching 100%), because it's a lot cheaper to create the illusion than the reality, and most consumers turn out to not have *that* much of a premium for the reality. And the reality of this lack of reality slowly corrodes away enthusiasm for it, eliminating that intangible's premium. A century later...
What examples do we have of genuinely superhuman performance aside from Chess? I'm not sure we have enough to generalize from. There are a variety of examples of historical automation with different dynamics (bank tellers, accountants upon the introduction of spreadsheet software).
I think it's hard to say in part because there is no human premium, so as soon as we attain truly superhuman performance on a whole task, not a single subpart, it is quickly defined away and becomes nothing but an amusing footnote in history texts about absurdly inefficient and wasteful ways to do something, if that.
Have you ever paid a 'knocker-upper' lately to wake you up? How many people have you hired to operate an old HP desk calculator for you of late? "No, of course not, Gwern, that would be ludicrous, I have an alarm on my smartphone and I have no use for a 'computer' to do arithmetic for me when I have digital computers. Why would I want to hire a human for any of that?" Yeah, exactly. You don't even realize you don't have a 'human premium' for countless tasks you take for granted that a machine always does now, even though they *didn't* always do them. You don't have a human premium for harvesting corn in Iowa by hand, either, or pretty much anything done by a machine.
Notice that in the only examples you gave, machines *don't* clearly have superhuman performance, because bank tellers and accountants can do a lot more than just a few fragmented tasks that ATMs and spreadsheets can do.
You’re basically arguing for the Butlerian Jihad right now.
(I personally would not because I don’t think it would work. See: Ukraine. Also, as a Dune fan, I must point out that the Butlerian Jihad is widely misunderstood, in large part due to the wretched sequels/prequels, and we face a scenario closer to the end of _Chapterhouse: Dune_, rather than the Butlerian Jihad.)
I meant more in the sense that if your arguments are true then the only ethical option would be a crusade against computers, not that this is what you were actually suggesting.
I also disagree with that. I don’t think it’s moral to do things which don’t work and make the world a worse place; I think it’s moral to do things which make the world a better place.
Yea I found it weird that both Ezra and Imas didn't really address this fairly straightforward point:
"Just because something is scarce and high-status doesn’t mean that everyone will be able to afford it; in fact, many people won’t."
We're assuming that most people will be able to afford all these highly priced relational goods on the basis of what...that more people will be in high-income relational jobs? Or because commodity prices will drop?
Reminds me of de-industrialization -- your community crumbled when the factory closed, but it's gonna be ok because that was offset by cheap tvs. There are going to be a lot of losers. And the richest 10% already make up over half of consumer spending. So this sounds right:
"So we may end up in a world where most people are handmaidens to the super-rich, while most normal people settle for cheap digital alternatives."
Thank you for helping me understand AI and the impact it will have on the future of work. The pragmatism China has demonstrated and the realisation that it is a tool that will impact society in massive ways.... I liked the argument that AI has cost determinants in China and can see how the argument will develop in the future. The argument reminds me of the birth of digital and the revolt against mechanisation in the nineteenth century cotton mills in the UK... there are so many parallels throughout history... It is a wonderful article... thank you....
1:1 teacher student ratio would be great for when student population drops with population to match the people who want to be teachers and are reasonably good at teaching, but education is a whole other can of worms without AI in the mix.
The full-time children feels like a compromise on traditional family dynamics--in the past, the eldest son or daughter's family might have been expected to care for the parents (depends on regional traditions), but sometimes a kid who doesn't marry might perform this role. In a family with many children who survive to adulthood, you might have some complex family inheritance wrangling mixed in for who gets to provide the care (for good or ill) so they control the parents' wealth. I have some experience with this from some older aunts/uncles who had many children (4-9) in Shanghai, though I also know some happy and drama-free families too.
Regarding jobs generally, there will always be headaches in how people see their wage based on endogenous factors that will just be different across many people. What do they want, what do they need, are there food, restaurant, leisure, clothing, whatever brands that they absolutely must have? What is their cost of living where they live? Maybe they can't move to a cheaper place because they don't have a car or a child or another dependent or partner can't move. Maybe they are stuck somewhere because of the peculiarities of tax or property laws we have on state and local levels in the US. A person happy with a multigenerational/multiple wage family in one house situation might be able to save a lot more than a single person living in an RV park. Remember that Jevon's Paradox came out of 1865... How people lived (larger families, people migrating to industrial areas may have had hometown migrants to help each other) and spent money (less leisure/goods, many things also more expensive) on was a lot different than where we are today (though it might feel familiar for many Chinese and others in developing countries).
Great piece! The demand-dampening point is what the Jevons optimists keep skating past. Cheaper production only expands demand if the wealth circulates broadly enough to generate it. If it concentrates, you hit a ceiling. The top 1% can't consume their way to replacing middle-class spending. China already shows this: massive productive capacity, weak domestic consumption, high savings rate. It's not a theoretical risk.
The doomsday prophecy of the "permanent underclass" is not something a vague "societal choice" can reverse. Silicon Valley is not "bracing" (what an egregious, disingenuous headline. Assuming NYT did it) but gleeful (its CEOs and owners def seem to be) of their own 'can't be helped' belt-tightening, unchecked power, speculation and prospects.
Long way of saying, I wish your piece had started with reversing the inevitability of these death knell "permanent underclass" proclamations (know you tried to later in the piece). Clearly, the smart researchers and scientists you spoke to over months, had practical ideas besides uncontrolled civil upheaval and the bad guys winning.
I agree with Jasmine. Wealth distribution will be required to appease the hungry. Roll on Socialism.
I pretty much agree with Jasmine. The irony for the AI originators is that the riches they make will require state regulation as in China to re-distribute among the out of work poor that AI will create. If not then no barricade will keep out a bored and hungry mob. If I had to make a prediction it would be that the next US government will likely be run by a Socialist President voted in by the jobless.
Your email thanks arrived. I'm also on Substack now.
I wrote a reply to your email noting we were in the same area of interest, since you asked. Substack wrote me this is strictly verboten.
My interest and work go back to 2002, but actually the 1970's where the world we now inhabit was being conceived without digital or cloud technology.
In any case, I will likely continue this work in China later this year.
super interesting as usual. I didn't realize you are a free-lancer. I hope you can make a livelihood and keep doing this. I'm really grateful for your work.