Agree on generative AI, at least from my sample of 1. I can never get it to turn out even boilerplate text in less time than it would take me to type something better. But it's comparable to my median student in writing quality, if you don't worry about its capacity for confabulation.
“The master craftspeople of software engineering, financial management, and business communication might find that with just a little help from an AI assistant, a bunch of normal people can do their job.”
Anecdotal but I’m seeing something like this in my professional life: a kind of democratization of code writing for After Effects motion graphics software. Stuff that non-coding motion designers would normally have to find (and often buy) they are starting to just getting GPT to write an expression solving the specific problem or effect they want....
Hi Noah, first time commenter here, but I’ve been reading for a little while, I really appreciate the newsletter.
I wonder whether the stories about employment and AI aren’t more complex than “is automation replacing jobs or not” - you seem to agree - and so maybe looking at the overall unemployment rate is too simplistic or even misleading. It seems to me like a lot of job growth is happening in jobs that are not “good jobs” - low salaries, no opportunities for advancement, no benefits, little on the job training, etc. I don’t know if the data backs this up. And if that’s true, it is probably due in some measure to automation/technology, and also globalization, monopoly power, etc.
For instance, ecommerce has grown tremendously and created lots of jobs in fulfillment centers, while perhaps leading to a decrease in jobs in physical retail. I suspect that fulfillment centers, due to the use of automation, are steeper hierarchies: fewer middle managers necessary, fewer specialists necessary, masses of manual workers. Perhaps physical retail had more middle management (per store), more specialists in merchandising, local marketing, etc. So overall increase in employment, but the labor market is transformed more into an hour glass.
Or another example with AI, even before the recent wave of generative AI: taxi transportation. Maybe this has been studied, but here is a sector in which the job of “dispatcher” was replaced by AI. But also local management of taxi companies, administrative staff etc. Probably an overall increase in employment, some hundreds or thousands of jobs created to engineer the algorithms at Uber HQ, plus lots of new drivers, and maybe an increase in consumer choice and satisfaction. But the labor market lost some jobs in the middle, and split into the mass of low paid jobs and the groups of elite workers at Uber.
I’d love to hear your take on this.
And it relates to the point you make at the end, about AI helping normies. In the example you mention, I wonder whether those customer service workers saw their pay increase along with the increased productivity. If AI makes people more productive, but they remain on the bottom of the hourglass distribution, then the implications are different than what we might naturally assume. When a checkout clerk at a grocery store can monitor 6 self-checkout stations at once, their productivity is higher, but I bet they make the same wage…
Noah, you put so much work into these long and thorough posts, then you can't resist throwing in a line about the mRNA vaccines "doing wonders for covid" to make me question how much to trust everything else? Even the imprecise phrasing, "doing wonders" sounds like a dog-whistle to (your audience? the Good People?) that you promise not to look into *that* topic too closely. You will dutifully Say The Thing.
First of all your framing is backwards. You make it sounds like the mRNAs did so well against covid that now they are expanding into cancer. I believe until quite recently, cancer was their most likely use case; they were never considered as a candidate for widespread vaccination of healthy populations because the side effect profiles were too bad for that use case. This was overcome only with an insanely powerful propoganda campaign that convinced people feeling miserable for 3 days meant "the vaccine was working." Now that it's clear (see bivalent uptake) that no more than ~20% of the population will voluntarily line up for an mRNA covid shot again, the return to cancer is a bit of a commercial retreat, not some validation.
But more generally, on what planet have they "worked wonders" against covid? The planet where the original 2 shots "ended the pandemic" like 90% of the population was led to believe they would (otherwise they would not have agreed to lockdown conditions for a year+ pending this miracle vacccine?) Or the actual planet where the current messaging is basically "your 6th booster might lessen symptoms for 3 months after taking it, but we're not sure?" There is decent evidence the vaccines may even have *negative* efficiency against omicron infection, a ridiculous thing for a vaccine to have. It's okay to grapple with the fact they are a disappointment. People have touched the third rail and survived.
Good point that US state capacity is higher than people make it out to be. I wish people were more attuned to good news and success stories like that, which is one of the reasons I appreciate your point of view.
I do wish we could make more progress in taming housing costs, investing in infrastructure and transportation, and improving education and healthcare. I'm hoping we see continued zoning and permitting reform and maybe the rise of AI-assisted self study to address some of those areas.
On poverty, I think the chart you used was based on the Supplemental Poverties Measure (SPM), which is a more ambitious attempt to track poverty, adjusting for geography, net transfers from govt., and other factors. The news is good on an absolute measure basis, but it is always necessary to look behind the numbers to assess where we are.
The SPM is based on (from Congressional Resource Service) :
"The SPM poverty thresholds measure a standard of living based on expenditures for food, clothing, shelter, and utilities (FCSU), and “a little more” for other expenses. The resources measured against those thresholds represent disposable income (after taxes and certain other expenses), including the value of certain noncash benefits, that are available to families to meet those needs."
The average SPM threshold is about $30,000 for a household of two adults and two children. . Or about $20 per day per family member.
Poverty should certainly be measured on an absolute basis. But, as well, we ought to measure poverty on a relative basis, e.g., how many households are below 50% of a country's median income. Because poverty is both absolute and relative, both to how others are doing and what a country is capable of providing.
Yesterday, I wrote a post (under 1,000 words) about the harm of exaggerated/false praise as it related to, among other things, tennis lessons, George Bush and most importantly global poverty. I was dismayed at what I found.
AI is a tool, not a menace. Just as the leverage principle and pulley sets permit relatively less powerful individuals to accomplish more work, the tool of AI increases the capability off intellectual labor. Of course told can be abused. The leverage principle permits you to do more damage with a club than a bare hand. Institutional constraints must be developed to prevent abuse. We don't need a generation of AI con artists scamming is.
Again, open the prisons and release non-violent offenders into training programs and jobs. It makes little sense to spend a five-figure sum housing each prisoner, when you can save money allowing them to enter the work force and become a “productive member of society.” Anything else is just lip service and justification of the world’s largest incarceration business.
Noah the stats on unemployment are fudged. Unemployment is low but labour force participation rates are at an extreme low. Think it’s a really important point you’re overlooking. Check out some of nicholas eberstaadts work in the area.
I would love to see a reader request feature. On that note, could you do a post on what higher ed/STEM education might look like in the future? I was struck in you "Big Chill" post at how many Millenials are pissed off about being sold false promises on the return to higher ed. But you've also written a lot (and re-posted today) about the importance of STEM education for future competitiveness. As someone working in STEM and STEM higher ed, I tend to agree with both of these points. But I haven't seen a compelling post on what this could look like other than "academic birth control", which would put many current graduate students and postdocs out of work, while doing nothing for maintaining scientific leadership. I'd really like to see a post on how science higher ed could be reformed to maintain or extend leadership in science while being less cruel to the participants.
Sorry, today I’m the friendly typo-finding guy guy:
“Also, there will be a wave of bank mergers, as smaller, weaker banks are absorbed by smaller ones.“
Absorbed by “larger” ones yes?
Agree on generative AI, at least from my sample of 1. I can never get it to turn out even boilerplate text in less time than it would take me to type something better. But it's comparable to my median student in writing quality, if you don't worry about its capacity for confabulation.
“The master craftspeople of software engineering, financial management, and business communication might find that with just a little help from an AI assistant, a bunch of normal people can do their job.”
Anecdotal but I’m seeing something like this in my professional life: a kind of democratization of code writing for After Effects motion graphics software. Stuff that non-coding motion designers would normally have to find (and often buy) they are starting to just getting GPT to write an expression solving the specific problem or effect they want....
(Also typo-making guy guy)
Hi Noah, first time commenter here, but I’ve been reading for a little while, I really appreciate the newsletter.
I wonder whether the stories about employment and AI aren’t more complex than “is automation replacing jobs or not” - you seem to agree - and so maybe looking at the overall unemployment rate is too simplistic or even misleading. It seems to me like a lot of job growth is happening in jobs that are not “good jobs” - low salaries, no opportunities for advancement, no benefits, little on the job training, etc. I don’t know if the data backs this up. And if that’s true, it is probably due in some measure to automation/technology, and also globalization, monopoly power, etc.
For instance, ecommerce has grown tremendously and created lots of jobs in fulfillment centers, while perhaps leading to a decrease in jobs in physical retail. I suspect that fulfillment centers, due to the use of automation, are steeper hierarchies: fewer middle managers necessary, fewer specialists necessary, masses of manual workers. Perhaps physical retail had more middle management (per store), more specialists in merchandising, local marketing, etc. So overall increase in employment, but the labor market is transformed more into an hour glass.
Or another example with AI, even before the recent wave of generative AI: taxi transportation. Maybe this has been studied, but here is a sector in which the job of “dispatcher” was replaced by AI. But also local management of taxi companies, administrative staff etc. Probably an overall increase in employment, some hundreds or thousands of jobs created to engineer the algorithms at Uber HQ, plus lots of new drivers, and maybe an increase in consumer choice and satisfaction. But the labor market lost some jobs in the middle, and split into the mass of low paid jobs and the groups of elite workers at Uber.
I’d love to hear your take on this.
And it relates to the point you make at the end, about AI helping normies. In the example you mention, I wonder whether those customer service workers saw their pay increase along with the increased productivity. If AI makes people more productive, but they remain on the bottom of the hourglass distribution, then the implications are different than what we might naturally assume. When a checkout clerk at a grocery store can monitor 6 self-checkout stations at once, their productivity is higher, but I bet they make the same wage…
Anyway, thanks for writing about all this!
Noah, you put so much work into these long and thorough posts, then you can't resist throwing in a line about the mRNA vaccines "doing wonders for covid" to make me question how much to trust everything else? Even the imprecise phrasing, "doing wonders" sounds like a dog-whistle to (your audience? the Good People?) that you promise not to look into *that* topic too closely. You will dutifully Say The Thing.
First of all your framing is backwards. You make it sounds like the mRNAs did so well against covid that now they are expanding into cancer. I believe until quite recently, cancer was their most likely use case; they were never considered as a candidate for widespread vaccination of healthy populations because the side effect profiles were too bad for that use case. This was overcome only with an insanely powerful propoganda campaign that convinced people feeling miserable for 3 days meant "the vaccine was working." Now that it's clear (see bivalent uptake) that no more than ~20% of the population will voluntarily line up for an mRNA covid shot again, the return to cancer is a bit of a commercial retreat, not some validation.
But more generally, on what planet have they "worked wonders" against covid? The planet where the original 2 shots "ended the pandemic" like 90% of the population was led to believe they would (otherwise they would not have agreed to lockdown conditions for a year+ pending this miracle vacccine?) Or the actual planet where the current messaging is basically "your 6th booster might lessen symptoms for 3 months after taking it, but we're not sure?" There is decent evidence the vaccines may even have *negative* efficiency against omicron infection, a ridiculous thing for a vaccine to have. It's okay to grapple with the fact they are a disappointment. People have touched the third rail and survived.
Good point that US state capacity is higher than people make it out to be. I wish people were more attuned to good news and success stories like that, which is one of the reasons I appreciate your point of view.
I do wish we could make more progress in taming housing costs, investing in infrastructure and transportation, and improving education and healthcare. I'm hoping we see continued zoning and permitting reform and maybe the rise of AI-assisted self study to address some of those areas.
On poverty, I think the chart you used was based on the Supplemental Poverties Measure (SPM), which is a more ambitious attempt to track poverty, adjusting for geography, net transfers from govt., and other factors. The news is good on an absolute measure basis, but it is always necessary to look behind the numbers to assess where we are.
The SPM is based on (from Congressional Resource Service) :
"The SPM poverty thresholds measure a standard of living based on expenditures for food, clothing, shelter, and utilities (FCSU), and “a little more” for other expenses. The resources measured against those thresholds represent disposable income (after taxes and certain other expenses), including the value of certain noncash benefits, that are available to families to meet those needs."
The average SPM threshold is about $30,000 for a household of two adults and two children. . Or about $20 per day per family member.
Poverty should certainly be measured on an absolute basis. But, as well, we ought to measure poverty on a relative basis, e.g., how many households are below 50% of a country's median income. Because poverty is both absolute and relative, both to how others are doing and what a country is capable of providing.
Yesterday, I wrote a post (under 1,000 words) about the harm of exaggerated/false praise as it related to, among other things, tennis lessons, George Bush and most importantly global poverty. I was dismayed at what I found.
https://robertsdavidn.substack.com/p/exaggerated-or-false-praise-is-often
The War on Poverty wasn't lost. The government changed sides.
If the banks want their depositors to come back or to attract new ones how about increasing the interest they pay on deposits a bit
My man is leaning into the listicle
AI is a tool, not a menace. Just as the leverage principle and pulley sets permit relatively less powerful individuals to accomplish more work, the tool of AI increases the capability off intellectual labor. Of course told can be abused. The leverage principle permits you to do more damage with a club than a bare hand. Institutional constraints must be developed to prevent abuse. We don't need a generation of AI con artists scamming is.
Again, open the prisons and release non-violent offenders into training programs and jobs. It makes little sense to spend a five-figure sum housing each prisoner, when you can save money allowing them to enter the work force and become a “productive member of society.” Anything else is just lip service and justification of the world’s largest incarceration business.
Is this the official face reveal?
Noah the stats on unemployment are fudged. Unemployment is low but labour force participation rates are at an extreme low. Think it’s a really important point you’re overlooking. Check out some of nicholas eberstaadts work in the area.
Here's to internet fragmentation!
I would love to see a reader request feature. On that note, could you do a post on what higher ed/STEM education might look like in the future? I was struck in you "Big Chill" post at how many Millenials are pissed off about being sold false promises on the return to higher ed. But you've also written a lot (and re-posted today) about the importance of STEM education for future competitiveness. As someone working in STEM and STEM higher ed, I tend to agree with both of these points. But I haven't seen a compelling post on what this could look like other than "academic birth control", which would put many current graduate students and postdocs out of work, while doing nothing for maintaining scientific leadership. I'd really like to see a post on how science higher ed could be reformed to maintain or extend leadership in science while being less cruel to the participants.