Apologies but this is utterly ridiculous. There is such vicious opposition to even the very modest redistribution that the US does.
"Redistribution might be expensive, but the same AI-driven economic growth that generated the rising inequality would also create the fiscal space needed to offset it. In short, spreading income around is a political challenge, not a policy or economic challenge.
The issue here is that you're almost certainly not 150 years old, so in your working lifetime, you have only seen more and more inequality instead of more and more redistribution in the US. But the US has indeed seen a move from a very unequal society with a high GINI coefficient to a more equal society with a lower GINI coefficient before, between the Gilded Age and the 1970's, so such a move (again) can not be ruled out.
I stopped at the musician analogy. Since 1900 the US population has increased over four and a half times, but, according to this chart, the number of employed musicians has only increased two and a half times. On a per capita basis the number of employed musicians is about half of what it was in 1900. Not a reason for optimism.
The US also got many times wealthier too, absolute and per capita.
By the way, the post makes much of how demand for live musicians is 'normal good', but given the drop in consumption of live musicians by each consumer, wouldn't that make them actually an 'inferior good'? ("In economics, a normal good is a type of a good for which consumers increase their demand due to an increase in income, unlike inferior goods, for which the opposite is observed. When there is an increase in a person's income, for example due to a wage rise, a good for which the demand rises due to the wage increase, is referred as a normal good." https://en.wikipedia.org/wiki/Normal_good )
Yes, I believe you are correct. And, the obvious reason is the proliferation of music-on-demand, whether recordings or streaming. However, people at the highest end of the income scale do seem to be increasing consumption of live music via house concerts and private events. My musician friends say that this can make the difference in allowing them to be full-time musicians.
They could be a Veblen good at some point, but I don't know. If we were to make historical comparisons, often rich people would have fulltime musician employees, like Bach. I cannot, off the top of my head, name a single billionaire who keeps a personal composer or choirmaster, say, the way aristocrats used to do routinely. (I have read often about top musicians doing personal birthdays or parties for eyewatering sums, but even those are generally only as one-off gigs, and so fit easily into the view of ever-decreasing relative employment with skewed superstar dynamics.)
I think at all points live music has been a normal good, but there have been several technology shocks to the demand for live music. It's not like rich people don't pay more for concerts today than poor people, it's that other things happened during the time that the whole country got richer.
To take a more extreme example, horses were a normal good in 1880 and a normal good in 2026 but there are a lot fewer horses now.
I don't follow. If people got richer but bought a lot less live music, then how is that a normal good ("a type of a good for which consumers *increase* their demand due to an increase in income,") and not an inferior good? And similarly for horses? How exactly are they normal goods and not inferior goods?
Because at every time point, demand for live music/horses increased with income. The issue is that you can't reason about changes in demand due to income based on time series data because other things happen besides income changes.
OK, but that doesn't help the argument here. If your argument is that "actually, they are a normal good, they just act like an inferior good because of technological changes which happen to substitute or change what that good used to provide", well, since we're discussing technological changes here focused on human labor and cognition... So it seems someone arguing from musicians has a dilemma: either it's an inferior good straightforwardly and so the entire argument collapses from the beginning, or it's a normal good but it's acting inferiorly for exactly the same reasons that humans would in a full AGI scenario and the real world examples of musicians/horses help demonstrate why 'normality' doesn't save you from substitution.
After that, there's a defensible 'plutonomics' claim that there may be a small residual absolute demand by plutocrats for a bit of specialized human labor, in the way that billionaires might once in a while hire a string quartet or a particularly famous musician they like for a party or a rich woman likes to play at dressage, but the Panglossian views definitely do not follow that this demand will be anywhere what it is now, support more than a tiny fraction of humanity, last indefinitely, etc. (Especially as such views continue to punt on the question of why you expect those billionaires to be human or maintain their property rights or maintain a taste for humanity, and that sort of thing...)
The only claim I've made here is that live music is a normal good; that is, in any given situation, if someone had more money they would purchase more live music. It's obviously true that being a normal good doesn't save something from substitution -- for example, Swiss watches are a normal good but substitution hit them hard.
The conclusion from the point about musicians is not "AI will never replace any jobs that are done in person"; that would be silly. It's that people have a strong desire for human contact and if they're really rich they will pay for that.
Obviously all bets are off if actually almost everyone is impoverished in the AI future but that's a different claim than that we'll have a Star Trek future where everything is provided by AIs.
I found the lack of normalized numbers puzzling as well. Yes, the number of musicians went up, but what is it per capita? How was inflation-adjusted median income of musicians affected? What proportion of the economic activity was even attributed to musicians (or other jobs that needed "human touch")? The anecdotal jobs mentioned here – personal touch needed in music, sales job, service jobs, etc. – doesn't make me hopeful for a very bright future. Rather, it creates a vague picture of some passable future.
I mean, several generations ago, most (men) were doing dangerous mind numbing factory work (at low pay too before mass unionization) and before that, it was backbreaking dangerous farm work, so . . . .
I like this take. It’s easy to forget that automation has been “good enough” to replace many roles for decades, yet demand for real humans still persists.
I do wonder whether the “human touch as a normal good” eventually becomes stratified. As incomes rise, people may increasingly pay a premium for human-delivered experiences, while AI becomes the default at the lower end. That would suggest AI doesn’t eliminate work so much as re-segment it around trust, presence, and signal.
Yes, that is likely what will happen. The poor will get robot nurses and doctors. The well off will get human nurses and doctors (to at least provide a human face/facade to the diagnosis provided by AI).
Maybe. Or maybe professions with political power will act to gatekeep access to AI advice, like Sudanese passport control officers.
AI already provides better advice and treatment recommendations for many conditions. Reportedly it can do badly with edge cases, but the papers trumpeting that finding carefully do not discuss iatrogenic harm rates from median human professionals. Nor do they discuss accessibility for median citizens.
I can't really blame the professions. They don't want to slip from the middle class into the masses and they will use whatever power they have to hold on to their position. Most people having to suffer poor health as a result is somebody else's problem.
Great piece and I fully second it. The classic example of the ATM comes to mind as well. I think the danger to work is over-stated. I actually think the genAI can lead us to focus on what's truly 'value add'. The concerns around entry-level white collar jobs are good example of it - the purpose of the job was not just making a slide deck, which it feels like it morphed into.
Classic "first we assume a can opener" economist take. At least there is an implicit admission that this is fundamentally a political problem, but there are no examples of what redistributive policy should look like.
There will be jobs, but they will all essentially be "be a status object to show my wealth and power". Like the domestic servants and palace artists and philosophers of old.
I recommend trophy wife if you want that kind of work. The working conditions are better than average.
"On paper, the job of waiter has been fully automated for over a decade. And yet, today there remain 1.9 million waiters across the US."
----
That's because robots are not yet advanced enough to do the tasks that waiters are required to do. Check back in a year or so.
People will be more than happy to save 20% or more on their meals when they don't have to tip robot workers. They will probably eat out more often!
The vast majority of people are not going to care whether a human made something or not. Do you care who made your car or TV or computer? The answer is no.
"But it is essential to balance the discussion with some optimism. I can see glimmers of hope in a simple fact: There are many jobs and tasks that easily could have been automated by now — the technology to automate them has long existed — and yet we humans continue to do them. The reason is that demand will always exist for certain jobs that offer what I call “the human touch.”"
-----
That's pretty specious logic! As AI continues to improve and proves that it can do any work better than humans, then that work WILL go to the machine.
In order for humans to keep their job in this situation, they would have to produce at least equal results to the machine AND they would have to do so faster and cheaper than the machines. Do you REALLY think that is going to be possible?
Most human work is done in a "good enough" manner. It does not have to be perfectly done, it just has to be "good enough". Machines will certainly be able to do better.
I read the waiter analogy. Actually payments is technically the least human action in the world. Nobody likes to ask for money and very few like to be asked for money. Big success in Uber and Airbnb is that the whole payment part was automated. In restaurants automating payments should be mandatory. So the waiter can focus on service.
McDonalds automated order taking and payments, yet they increased headcount because now they were offering table service.
Apologies but this is utterly ridiculous. There is such vicious opposition to even the very modest redistribution that the US does.
"Redistribution might be expensive, but the same AI-driven economic growth that generated the rising inequality would also create the fiscal space needed to offset it. In short, spreading income around is a political challenge, not a policy or economic challenge.
The issue here is that you're almost certainly not 150 years old, so in your working lifetime, you have only seen more and more inequality instead of more and more redistribution in the US. But the US has indeed seen a move from a very unequal society with a high GINI coefficient to a more equal society with a lower GINI coefficient before, between the Gilded Age and the 1970's, so such a move (again) can not be ruled out.
I stopped at the musician analogy. Since 1900 the US population has increased over four and a half times, but, according to this chart, the number of employed musicians has only increased two and a half times. On a per capita basis the number of employed musicians is about half of what it was in 1900. Not a reason for optimism.
The US also got many times wealthier too, absolute and per capita.
By the way, the post makes much of how demand for live musicians is 'normal good', but given the drop in consumption of live musicians by each consumer, wouldn't that make them actually an 'inferior good'? ("In economics, a normal good is a type of a good for which consumers increase their demand due to an increase in income, unlike inferior goods, for which the opposite is observed. When there is an increase in a person's income, for example due to a wage rise, a good for which the demand rises due to the wage increase, is referred as a normal good." https://en.wikipedia.org/wiki/Normal_good )
Yes, I believe you are correct. And, the obvious reason is the proliferation of music-on-demand, whether recordings or streaming. However, people at the highest end of the income scale do seem to be increasing consumption of live music via house concerts and private events. My musician friends say that this can make the difference in allowing them to be full-time musicians.
They could be a Veblen good at some point, but I don't know. If we were to make historical comparisons, often rich people would have fulltime musician employees, like Bach. I cannot, off the top of my head, name a single billionaire who keeps a personal composer or choirmaster, say, the way aristocrats used to do routinely. (I have read often about top musicians doing personal birthdays or parties for eyewatering sums, but even those are generally only as one-off gigs, and so fit easily into the view of ever-decreasing relative employment with skewed superstar dynamics.)
I think at all points live music has been a normal good, but there have been several technology shocks to the demand for live music. It's not like rich people don't pay more for concerts today than poor people, it's that other things happened during the time that the whole country got richer.
To take a more extreme example, horses were a normal good in 1880 and a normal good in 2026 but there are a lot fewer horses now.
I don't follow. If people got richer but bought a lot less live music, then how is that a normal good ("a type of a good for which consumers *increase* their demand due to an increase in income,") and not an inferior good? And similarly for horses? How exactly are they normal goods and not inferior goods?
Because at every time point, demand for live music/horses increased with income. The issue is that you can't reason about changes in demand due to income based on time series data because other things happen besides income changes.
OK, but that doesn't help the argument here. If your argument is that "actually, they are a normal good, they just act like an inferior good because of technological changes which happen to substitute or change what that good used to provide", well, since we're discussing technological changes here focused on human labor and cognition... So it seems someone arguing from musicians has a dilemma: either it's an inferior good straightforwardly and so the entire argument collapses from the beginning, or it's a normal good but it's acting inferiorly for exactly the same reasons that humans would in a full AGI scenario and the real world examples of musicians/horses help demonstrate why 'normality' doesn't save you from substitution.
After that, there's a defensible 'plutonomics' claim that there may be a small residual absolute demand by plutocrats for a bit of specialized human labor, in the way that billionaires might once in a while hire a string quartet or a particularly famous musician they like for a party or a rich woman likes to play at dressage, but the Panglossian views definitely do not follow that this demand will be anywhere what it is now, support more than a tiny fraction of humanity, last indefinitely, etc. (Especially as such views continue to punt on the question of why you expect those billionaires to be human or maintain their property rights or maintain a taste for humanity, and that sort of thing...)
The only claim I've made here is that live music is a normal good; that is, in any given situation, if someone had more money they would purchase more live music. It's obviously true that being a normal good doesn't save something from substitution -- for example, Swiss watches are a normal good but substitution hit them hard.
The conclusion from the point about musicians is not "AI will never replace any jobs that are done in person"; that would be silly. It's that people have a strong desire for human contact and if they're really rich they will pay for that.
Obviously all bets are off if actually almost everyone is impoverished in the AI future but that's a different claim than that we'll have a Star Trek future where everything is provided by AIs.
I found the lack of normalized numbers puzzling as well. Yes, the number of musicians went up, but what is it per capita? How was inflation-adjusted median income of musicians affected? What proportion of the economic activity was even attributed to musicians (or other jobs that needed "human touch")? The anecdotal jobs mentioned here – personal touch needed in music, sales job, service jobs, etc. – doesn't make me hopeful for a very bright future. Rather, it creates a vague picture of some passable future.
Travel agents, wait staff, sales people and insurance agents? Sounds like a bright AI future for the American worker.
I mean, several generations ago, most (men) were doing dangerous mind numbing factory work (at low pay too before mass unionization) and before that, it was backbreaking dangerous farm work, so . . . .
Funny how the only thing stopping the human touch from outcompeting porn, is that prostitution is illegal.
In fact the trend in that sphere is also towards human interaction, viz OnlyFans.
The person who wrote this needs to think harder...
I'm sure everyone here has read the Matt Shumer article. Noah Smith with a keen insight too: https://www.noahpinion.blog/p/you-are-no-longer-the-smartest-type
I like this take. It’s easy to forget that automation has been “good enough” to replace many roles for decades, yet demand for real humans still persists.
I do wonder whether the “human touch as a normal good” eventually becomes stratified. As incomes rise, people may increasingly pay a premium for human-delivered experiences, while AI becomes the default at the lower end. That would suggest AI doesn’t eliminate work so much as re-segment it around trust, presence, and signal.
Yes, that is likely what will happen. The poor will get robot nurses and doctors. The well off will get human nurses and doctors (to at least provide a human face/facade to the diagnosis provided by AI).
Maybe. Or maybe professions with political power will act to gatekeep access to AI advice, like Sudanese passport control officers.
AI already provides better advice and treatment recommendations for many conditions. Reportedly it can do badly with edge cases, but the papers trumpeting that finding carefully do not discuss iatrogenic harm rates from median human professionals. Nor do they discuss accessibility for median citizens.
I can't really blame the professions. They don't want to slip from the middle class into the masses and they will use whatever power they have to hold on to their position. Most people having to suffer poor health as a result is somebody else's problem.
Great piece and I fully second it. The classic example of the ATM comes to mind as well. I think the danger to work is over-stated. I actually think the genAI can lead us to focus on what's truly 'value add'. The concerns around entry-level white collar jobs are good example of it - the purpose of the job was not just making a slide deck, which it feels like it morphed into.
Redistribution.
Classic "first we assume a can opener" economist take. At least there is an implicit admission that this is fundamentally a political problem, but there are no examples of what redistributive policy should look like.
There will be jobs, but they will all essentially be "be a status object to show my wealth and power". Like the domestic servants and palace artists and philosophers of old.
I recommend trophy wife if you want that kind of work. The working conditions are better than average.
"On paper, the job of waiter has been fully automated for over a decade. And yet, today there remain 1.9 million waiters across the US."
----
That's because robots are not yet advanced enough to do the tasks that waiters are required to do. Check back in a year or so.
People will be more than happy to save 20% or more on their meals when they don't have to tip robot workers. They will probably eat out more often!
The vast majority of people are not going to care whether a human made something or not. Do you care who made your car or TV or computer? The answer is no.
"But it is essential to balance the discussion with some optimism. I can see glimmers of hope in a simple fact: There are many jobs and tasks that easily could have been automated by now — the technology to automate them has long existed — and yet we humans continue to do them. The reason is that demand will always exist for certain jobs that offer what I call “the human touch.”"
-----
That's pretty specious logic! As AI continues to improve and proves that it can do any work better than humans, then that work WILL go to the machine.
In order for humans to keep their job in this situation, they would have to produce at least equal results to the machine AND they would have to do so faster and cheaper than the machines. Do you REALLY think that is going to be possible?
Most human work is done in a "good enough" manner. It does not have to be perfectly done, it just has to be "good enough". Machines will certainly be able to do better.
I read the waiter analogy. Actually payments is technically the least human action in the world. Nobody likes to ask for money and very few like to be asked for money. Big success in Uber and Airbnb is that the whole payment part was automated. In restaurants automating payments should be mandatory. So the waiter can focus on service.
McDonalds automated order taking and payments, yet they increased headcount because now they were offering table service.