A "progressive" who can't think of anything better for humans to do than churn out widgets on an assembly line is, ipso facto, not "progressive" in any sense of the word as I would use it.
Automation frees up human capital – time and energy – which should, at the margin, lead to more opportunities for human flourishing. What's not progressive about that?
Why not? Work is just one aspect of peoples lives. Manufacturing hours were heading to about 36 hours prior to the great sell off to China, which was often 3 days of 12 hours. Giving 4 days for flourishing.
With very few exceptions there’s little “self-actualisation” at work.
I was glad to see this. The recent book Prediction Machines, by Agrawal, Gans, and Goldfarb, made a convincing argument, about AI rather than automation, that this won’t displace jobs in the way feared because human judgment is a complement to cheap prediction. AI creates new input and faster that has to be interpreted by humans and also allows humans to switch to more valued-added aspects of their jobs, much as ATM machines did for bank tellers, whose numbers did not decline. There may be similar effects involved with mechanical automation as with AI.
I do think there are some winners and losers and think it would be compelling to examine the largest episodes of job displacement in the past. The vast secretary pools of the 1960s through 1980s were reduced to a fraction after computers with word processors connected to printers became standard issue for office workers. I have read that the number of dock workers declined enormously as old-fashioned longshoreman work was replaced with container shipping, providing some validation to the concerns of US dock workers today (though perhaps most of these jobs would have been mechanized some other way no matter what). What happened to these vast pools of labor and what did they end up doing instead?
The longshoremen had a pretty powerful union, so they negotiated a deal allowing port automation in exchange for income protection. Moving from break bulk along the Manhattan waterfront to containers near the turnpike in New Jersey saved enough money to more than cover the deal. Nowadays, there are very few unions, so the companies tend to grab all the savings and leave the workers to scramble and flounder.
Back in the 60s when I was a kid we were still getting the tailwinds of the Bucky Fuller utopia era where we would soon have 25 hour workweeks due to automation. While I think that is possible if you are willing to stagnate technologically, it's human nature to want the next best thing and for that always need human labor (inventors, financers, manufacturing overseers, etc.). And of course there's always the argument that someone needs to build, program, and maintain the robots. Even if we could pull off the 25 hour workweek we'd need labor to provide the luxuries for all that freed up time---hotel workers, ski lift operators, home renovation companies (ie--the pandemic showed us how all that free time was even busier than usual), and such.
COVID and work from home revealed just how many typical office hours are just wasted time. One of the reason that managers want the workers back in the office is so they can enforce wasted time. We could have just as much production of goods and services with a lot fewer hours of work.
People seem to have done fine with the hour or two a day that working from home saves. Having a life may be a luxury, but it doesn't require hiring a lot of workers.
Point taken, although my wife has experienced something of the opposite ---she switched to be an online teacher and now she spends more time teaching/grading than she did in a classroom. Yuck!
Perhaps, but at least it is better _accounted for_ than "all the time you've spent at school is teaching", and seeing the data is the first step to meaningfully change it.
Thanks for this excellent post - it’s frustrating that people’s mindset is to romanticize difficult, low-paying, low-value-add jobs instead of embracing the need to innovate and compete globally.
Thanks Noah. I can’t speak to the economic impact of actually existing robots, but I do think the coming automation apocalypse is grossly exaggerated.
I keep reading about how lawyers are going to be replaced by AI. But the articles making that claim never explain HOW. It’s always something-something-machine-learning-mumble. The only solid task they ever mention is brief writing. Don’t make me laugh. I work tangentially but often with up and coming AI legal research products. It isn’t even remotely close - like you say, it’s pure science fiction speculation.
I’ve been a lawyer for 20 years. The only task computers have replaced lawyers in is document review. The nasty secret of law firms, though, is that you never needed lawyers to do that in the first place, and you wouldn’t need (and couldn’t have) computers take it over if the task hadn’t exploded in the last forty years because electronic communications made document review several orders of magnitude larger.
"People find new jobs to do. And their incomes generally rise as a result." Yup, there is always more to do than people to do it. And most people know what they'd like to do if they had more time to do it. If we could automate the picking and sorting of apples on our farm, then we'd have the time and resources to turn that unused second story space in our barn into a dining room and event center.
The typical apple orchard these days is much more efficient than a bunch of trees. In Washington State, at least, the trees are espaliered, spread out on frames like grape vines. This lets in more light to improve fruit production and makes pruning and picking easier since it can be done standing at ground level. Fruit management jobs are now more skilled. Workers have to know how to prune the trees properly.
Meanwhile, apples are seen as less profitable, so a lot of production is going into hops, grown at high density in sort of cages, and grapes for wine. The hops jobs have only been slightly upskilled, but the grape work operates at a whole different level.
P.S. Hope you get the time to build out that dining room and event center.
“In Sweden, if you ask a union leader, ‘Are you afraid of new technology?’ they will answer, ‘No, I’m afraid of old technology,’” says the Swedish minister for employment and integration, Ylva Johansson.
Free trade will be good for workers too, they said.
Economists sung about Free trade in their sleep. They said gains were worth it and the workers who lost their jobs could find employment elsewhere. "Don't be stupid", they said. "You don't understand comparative advantage", they said. But it's wasn't so black and white. And It's not black and white here either. 'Robots are good' vs 'Robots are bad' isn't the right framework here.
We should not mock peoples' fear. Workers have seen their own bargaining power dissipate over the past 30 years. They have been doing more for less as productivity outpaced earnings. the economy has become artificial and financialized. It's natural for workers to be weary and frightened of fundamental changes as it relates to their earnings.
Common people are the best data collectors - they see observations with their own eyes. They've seen how technology is weaponized in warehouses to track every breathe they take. They've experienced the decline in labor markets in places far remote from cities like Boston or San Francisco. We should not tell people they're just stupid.
Your worry is that fear itself will lead to irrational inaction: that the us will not embrace innovation that will help the american worker. That is legitament. And Yes, there are instances that automation will help the average worker become more efficient. But that doesn't mean as a whole, automation will help people people more productive.
In my view, the economy has almost become too productive for regular people. We've created an hyper efficient and financialized economy in which we have decreased the inputs necessary to produce output.
Automation has already been occurring. If we define automation as a system that performs without the need or use of a person's actions, then automation has already been occurring for decades as companies have found ways to cut costs and decrease payroll, and the effects of this automation has just increased the top 10% power over the American people. Have people been benefiting? Are people enjoying the wondrous gains of innovation? Do people feel productive in their life and work? Do people have creative satisfaction in the work they do? Do people feel they have more bargaining power? Not at all.
We can dream of a world in which people use automation to make their work more productive. But really, automation will just be used to decrease worker inputs. Automation is coming for knowledge workers too. And my guess is when companies begin decreasing the payrolls for data scientists and computer programmers and accountants, then there will be a cultural change against singing the joys of innovation and progressive change. Already 'Learn to code' is thrown as a mocking insult, because programming itself has become automated. It's 100X harder to create value and get paid for it by coding today than it was in 2002.
I used to genuinely believe in your dedicated, thorough, level-headed, and data driven analysis of how the world works. But lately, I don't think I do anymore. Your writing has become too out-of-touch with people's experiences.
"I used to genuinely believe in your dedicated, thorough, level-headed, and data driven analysis of how the world works. But lately, I don't think I do anymore. Your writing has become too out-of-touch with people's experiences." <-- This just means that I started to say things that you didn't already agree with.
Man, did you not see the evidence in the post? Did you not see how employment has soared despite all the decades of automation? How wages have increased more strongly for those at the bottom for a decade now? Median personal incomes are up too.
Did you not read EPI's report? Are there any public intellectuals more dedicated to improving worker bargaining power and wages than the Economic Policy Institute? If so, I don't know who those would be. And EPI says that the fear of automation is misplaced.
Just because you feel threatened by something doesn't mean it's really a threat. That is why I try to provide dedicated, thorough, level-headed, data driven analyses of how the world works!
One thing to consider is how this argument played out when Britain industrialized its cloth industry. Yes, it created new jobs and new sources of wealth, but it totally destroyed the livelihoods of a huge cohort of workers. One could earn a good living producing cloth in the late 18th century, but you were out of work or working for pennies in the early 19th. Overall, society benefited dramatically, but that didn't help the displaced, only their grand and great-grand children.
Big economic transitions throw people to the side. When the nation moved its manufacturing south and then offshore, it was sold as a big win for everyone, but it was not a big win for the midwest which is still full of people who remember the many fine promises and saw their hopes for a decent living snatched away.
The problem with statistics is that one does not live a statistical life.
The evidence of employment “soaring” was 82% of people (at max) employed from the ages of 25-55 since the Reagan years. Soaring would be it going higher.
Perhaps go further back, since automation took off in earnest in the 60s or so.
"Your writing has become too out-of-touch with people's experiences." <-- This just means that I started to say things that you didn't already agree with."
No, it just means that he has learned more from greater exposure to your thinking and writing. Your absolute certainty, your tone of omnisciense, your lack of humility in ackowledging that there may be alternatives to your opinions and analyses of data are all things that become clearer with more exposure. There are shades of grey to everything - you don't acknowledge that in your writings. In short, you are what we used to call a "know-it-all".
Beatifully expressed and absolutely right. And the "winners and losers" argument is not well addressed in this article. In fact many consumers may benefit from lower prices due to lower cost from automation, while displaced workers are left stranded. That happens despite political promises of "retraining" which doesn't happen. Looking backward for evidence of automation not reducing jobs or lowering pay is inaccurate for at least two reasons: first, the workforce has failed to manintain its income levels relative to the growth in overall income (i.e. the rich get richer) and second, automation is still in its infancy and the effects haven't show up enough to be accurately measured by past data. And arguing from data from other counties that are more automated is specious because they often have, in fact, lower wage levels and/or have better adapted to the needs of displaced workers than the US. (i.e.subsiding workers though better welfare and minimum wage programs and shorter workweeks and better benefits.
There always are - that’s life. But we don’t need to celebrate/embrace it. And we should - and could - do more to help the losers than the lip serve we now offer.
I completely disagree with opinion of the article and your assertions. This is a very good convincing article by Noah with lots of evidence and addresses what you claim it doesn’t. As far as automation affecting my job ( software engineer, or coding as you called it) . When I started in high school in 1976 coding consisted of pencilling in dots on cards, putting the deck in a reader and waiting for a paper printout to get the result. Very slow and unproductive. Since then the automation applied to this job went from console terminals (type in the code and see results directly), disk drives for storage, text editors ( vi vs. EMACS), interactive debugging and integrated development environment apps), higher level languages, open source software libraries, source version control like GIT, automatic syntax checking and error correction, automated testing and performance analysis and now moving to machine learning based “code-less”programming. This was all dependent on progress in semiconductors and other computer technology , human interface design. If someone tried to create modern software (say the app you are reading this on without the automation developed since 1976, it would be as difficult as it would be to build a 747 with only hand tools and paper drawings. I’ve worked at software for 50 years including at Apple and Microsoft and I can say that while the hours spent working never went down, the quality, amount and functionality of shipped software increased by orders of magnitude. Unfortunately my salary didn’t increase by the nearly same amount, but productivity did and I was able to capture some of that by investing in my employers’ stock. At no time did my job feel in danger of being lost to automation.
With github and python I can see programming moving closer and closer to natural language until at some point the technical skill required for fixing problems as they occur becomes something my six year old can figure out. I remember when I started and their was a big learning curve and you had to buy books on C++. Now you can just look at the library definitions or google for an answer. Some day soon this won't be required because every branch of the etl will be indexed with quality control systems that bring up the broken part so that a person of modest intellect doesn't have to search, can see why it broke and then try a few things to come up with a fix. Their are companies out their trying to make data searchable like a google search. Eventually that will become more like a siri. We aren't their yet but in 40 years we will be.
A very good review. I recall a few decades ago my area was populated with dozens of textile mills making everything a person might wear. It employed thousands of local workers doing all sorts of jobs from starting floor workers to management. They paid well, they offered benefits, they gave people a way of life. In less than a 10-year period these jobs were moved out of the U.S. to China and Mexico, and other countries where textile companies could produce for less. They prided themselves on helping developing countries develop.
Today with automation and AI these types of jobs can and should return to the U.S. to create more jobs at levels never before seen. Less floor workers and more advanced skills in mechanization, maintenance, computer integration, and management.
The robots are available, many from China. I say buy them up, even give tax or other incentives to buy them. Doing so could create a tsunami wave of new growth and move the U.S. back to a self manufactured country rather than a net buyer from everyone else. It's our loss of market direction in this area that fuels the countries who today would want to see us decline.
Fascinating. It shows how difficult it is to predict trends. The exception to that might be demographics, and it would be interesting to see how robotics relates to that.
Really great article! There are lots of great reasons to bring manufacturing back (as you mentioned inflation reduction, supply chain bottlenecks, inequality) but I wanted your opinion on the risks associated with this as well. I imagine short term the benefits would outweigh but long term what about demand shifts to experiences rather than goods? Will millennials still prefer vacations to new cars long term and will that affect strategy for re-shoring manufacturing?
The issue is not robotic automation. It is robotic automation plus energy independence. The world is further along (perhaps much further along) on the path to accomplishing the first. When it accomplishes both, work will change completely.
Human labor can be divided into several components. Three primary ones are calculation, innovation, and mechanism. If they are given the same inputs and are asked to do the same calculation, our machines are much better than we are at calculation.
Innovation can be broadly thought of as ideation, pattern recognition, and error recognition. We are still doing better than the machines here. The goal of AI is to change that.
Much of the world's work, however, still involves the expenditure of human energy. Automation picks off those tasks as it becomes financially feasible to do so. Feasibility boils down to two things: Our ability to design replacement systems, and the cost of building them. We are getting better at the first every day. The second is a direct result of the cost of energy. Energy to power the machines that produce and modify the materials needed to build more machines.
Classical economic theory states that when a job is eliminated, the person holding that job moves on to what is now their highest and best use (i.e. the option that maximizes their utility). Two things need to be noted. First, this is not a guarantee that the new job will be better than the old one. In general, when all resources move to their highest use, society as a whole benefits. Individuals may not. Second, there is a floor below which it makes no sense for someone to accept their new roll. At this point, they will feel that they are better off going on the dole or turning to crime. Not everyone will be useful when most jobs involve programming.
It is hard to argue with any of the observations or conclusions in this article. The US certainly should not turn its back on any helpful technology, and being the best at important ones has got to be better than the alternative. But the best argument for using robots today is this: If you need to manage a tiger, it is best to start working with it when it is a cub. We need to consider what the world will look like in ten or 20 years.
> Not everyone will be useful when most jobs involve programming.
Programming will be automated long before nurse or occupational therapist or speech pathologist.
I think basically everyone will end up working in a human-human service role, for which the psychological requirements of the service basically by definition will exclude robots.
I am a software engineer of 10 years. Watching GitHub copilot’s progress in the last 14 months has me thinking my profession (as we know it, today) will be dead in 10 years.
Aren't economies dominated by human-to-human service roles historically notoriously terrible and exploitative ones reflective of a massive overproduction of humans relative to capital capable of productively employing them? Countries known for widespread use of domestic servants are, unsusprisingly, also known for attendant brutal labor exploitation and callous disregard for human rights. Being a domestic servant may be better than starving, but the widespread use and employment thereof seems like a much worse social equilibrium than the one we have now.
Obviously one rejoinder is that human to human service roles aren't limited to domestic servitude, but of course based on the remuneration to such roles in the present day -- whether daycare workers, home health aides, actors, or artists -- it would seem that we already have a relative glut of available labor even without massive displacement by capital, and that's notwithstanding the fact that at least daycare workers and home health aides are performing jobs with tremendous responsibility and value-add!
Yes, here surgeon is closer to software engineer than nurse. As you say, you’re not looking for human-human contact from a surgeon, just their steady hand.
"Our workers" can win against whom? Chinese workers? They're not the ones stealing from us, cheating us, usuring us, not paying us the value of our labor, keeping our wages low, our hours long, and our rents high. You're right that automation isn't the problem, but neither is China. The problem is capital.
Rentiers are often also capitalists, but it's the rent that's the problem. Georgist land value tax is the solution, and it could fund UBI which would give workers a much better bargaining position.
What's been described as "technological unemployment" is really just rentier capitalism at its core. The solution certainly isn't to smash up the robots and go full Hosni Mubarak on the Internet. Rather, it's to more fairly redistribute the robots and the Internet - including but not limited to breaking up Big Tech monopolies - so that they serve the public instead of dominating it.
It's very easy to forget that the original Luddites didn't sabotage weaving looms out of technophobia, they simply did so as a bargaining tactic for better pay and conditions, typically being the only ones who could get the looms moving again - a forerunner of the modern labour/trade union movement.
Excellent. Especially use of stats and graphs to build argument.
Thanks!!
A "progressive" who can't think of anything better for humans to do than churn out widgets on an assembly line is, ipso facto, not "progressive" in any sense of the word as I would use it.
Automation frees up human capital – time and energy – which should, at the margin, lead to more opportunities for human flourishing. What's not progressive about that?
Why not? Work is just one aspect of peoples lives. Manufacturing hours were heading to about 36 hours prior to the great sell off to China, which was often 3 days of 12 hours. Giving 4 days for flourishing.
With very few exceptions there’s little “self-actualisation” at work.
I was glad to see this. The recent book Prediction Machines, by Agrawal, Gans, and Goldfarb, made a convincing argument, about AI rather than automation, that this won’t displace jobs in the way feared because human judgment is a complement to cheap prediction. AI creates new input and faster that has to be interpreted by humans and also allows humans to switch to more valued-added aspects of their jobs, much as ATM machines did for bank tellers, whose numbers did not decline. There may be similar effects involved with mechanical automation as with AI.
I do think there are some winners and losers and think it would be compelling to examine the largest episodes of job displacement in the past. The vast secretary pools of the 1960s through 1980s were reduced to a fraction after computers with word processors connected to printers became standard issue for office workers. I have read that the number of dock workers declined enormously as old-fashioned longshoreman work was replaced with container shipping, providing some validation to the concerns of US dock workers today (though perhaps most of these jobs would have been mechanized some other way no matter what). What happened to these vast pools of labor and what did they end up doing instead?
The longshoremen had a pretty powerful union, so they negotiated a deal allowing port automation in exchange for income protection. Moving from break bulk along the Manhattan waterfront to containers near the turnpike in New Jersey saved enough money to more than cover the deal. Nowadays, there are very few unions, so the companies tend to grab all the savings and leave the workers to scramble and flounder.
Back in the 60s when I was a kid we were still getting the tailwinds of the Bucky Fuller utopia era where we would soon have 25 hour workweeks due to automation. While I think that is possible if you are willing to stagnate technologically, it's human nature to want the next best thing and for that always need human labor (inventors, financers, manufacturing overseers, etc.). And of course there's always the argument that someone needs to build, program, and maintain the robots. Even if we could pull off the 25 hour workweek we'd need labor to provide the luxuries for all that freed up time---hotel workers, ski lift operators, home renovation companies (ie--the pandemic showed us how all that free time was even busier than usual), and such.
Instead, we've chosen the worst of both - stagnation AND long hours. Great.
COVID and work from home revealed just how many typical office hours are just wasted time. One of the reason that managers want the workers back in the office is so they can enforce wasted time. We could have just as much production of goods and services with a lot fewer hours of work.
People seem to have done fine with the hour or two a day that working from home saves. Having a life may be a luxury, but it doesn't require hiring a lot of workers.
Point taken, although my wife has experienced something of the opposite ---she switched to be an online teacher and now she spends more time teaching/grading than she did in a classroom. Yuck!
Perhaps, but at least it is better _accounted for_ than "all the time you've spent at school is teaching", and seeing the data is the first step to meaningfully change it.
Manufacturing was heading to 36 hours across 3 days in many places. Which isn’t a great reduction in hours but a big increase in time off.
Thanks for this excellent post - it’s frustrating that people’s mindset is to romanticize difficult, low-paying, low-value-add jobs instead of embracing the need to innovate and compete globally.
Thanks Noah. I can’t speak to the economic impact of actually existing robots, but I do think the coming automation apocalypse is grossly exaggerated.
I keep reading about how lawyers are going to be replaced by AI. But the articles making that claim never explain HOW. It’s always something-something-machine-learning-mumble. The only solid task they ever mention is brief writing. Don’t make me laugh. I work tangentially but often with up and coming AI legal research products. It isn’t even remotely close - like you say, it’s pure science fiction speculation.
I’ve been a lawyer for 20 years. The only task computers have replaced lawyers in is document review. The nasty secret of law firms, though, is that you never needed lawyers to do that in the first place, and you wouldn’t need (and couldn’t have) computers take it over if the task hadn’t exploded in the last forty years because electronic communications made document review several orders of magnitude larger.
"People find new jobs to do. And their incomes generally rise as a result." Yup, there is always more to do than people to do it. And most people know what they'd like to do if they had more time to do it. If we could automate the picking and sorting of apples on our farm, then we'd have the time and resources to turn that unused second story space in our barn into a dining room and event center.
The typical apple orchard these days is much more efficient than a bunch of trees. In Washington State, at least, the trees are espaliered, spread out on frames like grape vines. This lets in more light to improve fruit production and makes pruning and picking easier since it can be done standing at ground level. Fruit management jobs are now more skilled. Workers have to know how to prune the trees properly.
Meanwhile, apples are seen as less profitable, so a lot of production is going into hops, grown at high density in sort of cages, and grapes for wine. The hops jobs have only been slightly upskilled, but the grape work operates at a whole different level.
P.S. Hope you get the time to build out that dining room and event center.
“In Sweden, if you ask a union leader, ‘Are you afraid of new technology?’ they will answer, ‘No, I’m afraid of old technology,’” says the Swedish minister for employment and integration, Ylva Johansson.
https://www.nytimes.com/2017/12/27/business/the-robots-are-coming-and-sweden-is-fine.html
Nice article!
A unionised workforce is a whole different thing.
Free trade will be good for workers too, they said.
Economists sung about Free trade in their sleep. They said gains were worth it and the workers who lost their jobs could find employment elsewhere. "Don't be stupid", they said. "You don't understand comparative advantage", they said. But it's wasn't so black and white. And It's not black and white here either. 'Robots are good' vs 'Robots are bad' isn't the right framework here.
We should not mock peoples' fear. Workers have seen their own bargaining power dissipate over the past 30 years. They have been doing more for less as productivity outpaced earnings. the economy has become artificial and financialized. It's natural for workers to be weary and frightened of fundamental changes as it relates to their earnings.
Common people are the best data collectors - they see observations with their own eyes. They've seen how technology is weaponized in warehouses to track every breathe they take. They've experienced the decline in labor markets in places far remote from cities like Boston or San Francisco. We should not tell people they're just stupid.
Your worry is that fear itself will lead to irrational inaction: that the us will not embrace innovation that will help the american worker. That is legitament. And Yes, there are instances that automation will help the average worker become more efficient. But that doesn't mean as a whole, automation will help people people more productive.
In my view, the economy has almost become too productive for regular people. We've created an hyper efficient and financialized economy in which we have decreased the inputs necessary to produce output.
Automation has already been occurring. If we define automation as a system that performs without the need or use of a person's actions, then automation has already been occurring for decades as companies have found ways to cut costs and decrease payroll, and the effects of this automation has just increased the top 10% power over the American people. Have people been benefiting? Are people enjoying the wondrous gains of innovation? Do people feel productive in their life and work? Do people have creative satisfaction in the work they do? Do people feel they have more bargaining power? Not at all.
We can dream of a world in which people use automation to make their work more productive. But really, automation will just be used to decrease worker inputs. Automation is coming for knowledge workers too. And my guess is when companies begin decreasing the payrolls for data scientists and computer programmers and accountants, then there will be a cultural change against singing the joys of innovation and progressive change. Already 'Learn to code' is thrown as a mocking insult, because programming itself has become automated. It's 100X harder to create value and get paid for it by coding today than it was in 2002.
I used to genuinely believe in your dedicated, thorough, level-headed, and data driven analysis of how the world works. But lately, I don't think I do anymore. Your writing has become too out-of-touch with people's experiences.
"I used to genuinely believe in your dedicated, thorough, level-headed, and data driven analysis of how the world works. But lately, I don't think I do anymore. Your writing has become too out-of-touch with people's experiences." <-- This just means that I started to say things that you didn't already agree with.
Man, did you not see the evidence in the post? Did you not see how employment has soared despite all the decades of automation? How wages have increased more strongly for those at the bottom for a decade now? Median personal incomes are up too.
Did you not read EPI's report? Are there any public intellectuals more dedicated to improving worker bargaining power and wages than the Economic Policy Institute? If so, I don't know who those would be. And EPI says that the fear of automation is misplaced.
Just because you feel threatened by something doesn't mean it's really a threat. That is why I try to provide dedicated, thorough, level-headed, data driven analyses of how the world works!
One thing to consider is how this argument played out when Britain industrialized its cloth industry. Yes, it created new jobs and new sources of wealth, but it totally destroyed the livelihoods of a huge cohort of workers. One could earn a good living producing cloth in the late 18th century, but you were out of work or working for pennies in the early 19th. Overall, society benefited dramatically, but that didn't help the displaced, only their grand and great-grand children.
Big economic transitions throw people to the side. When the nation moved its manufacturing south and then offshore, it was sold as a big win for everyone, but it was not a big win for the midwest which is still full of people who remember the many fine promises and saw their hopes for a decent living snatched away.
The problem with statistics is that one does not live a statistical life.
The evidence of employment “soaring” was 82% of people (at max) employed from the ages of 25-55 since the Reagan years. Soaring would be it going higher.
Perhaps go further back, since automation took off in earnest in the 60s or so.
"Your writing has become too out-of-touch with people's experiences." <-- This just means that I started to say things that you didn't already agree with."
No, it just means that he has learned more from greater exposure to your thinking and writing. Your absolute certainty, your tone of omnisciense, your lack of humility in ackowledging that there may be alternatives to your opinions and analyses of data are all things that become clearer with more exposure. There are shades of grey to everything - you don't acknowledge that in your writings. In short, you are what we used to call a "know-it-all".
Beatifully expressed and absolutely right. And the "winners and losers" argument is not well addressed in this article. In fact many consumers may benefit from lower prices due to lower cost from automation, while displaced workers are left stranded. That happens despite political promises of "retraining" which doesn't happen. Looking backward for evidence of automation not reducing jobs or lowering pay is inaccurate for at least two reasons: first, the workforce has failed to manintain its income levels relative to the growth in overall income (i.e. the rich get richer) and second, automation is still in its infancy and the effects haven't show up enough to be accurately measured by past data. And arguing from data from other counties that are more automated is specious because they often have, in fact, lower wage levels and/or have better adapted to the needs of displaced workers than the US. (i.e.subsiding workers though better welfare and minimum wage programs and shorter workweeks and better benefits.
What people need to realize is that:
1. There are winners and losers from trying to freeze our economy in amber.
2. There are winners and losers from letting our industry get outcompeted by China's industry.
3. There are winners and losers from supply bottlenecks.
There are going to be winners and losers, period.
There always are - that’s life. But we don’t need to celebrate/embrace it. And we should - and could - do more to help the losers than the lip serve we now offer.
Nope. The evidence is in the news. :)
I completely disagree with opinion of the article and your assertions. This is a very good convincing article by Noah with lots of evidence and addresses what you claim it doesn’t. As far as automation affecting my job ( software engineer, or coding as you called it) . When I started in high school in 1976 coding consisted of pencilling in dots on cards, putting the deck in a reader and waiting for a paper printout to get the result. Very slow and unproductive. Since then the automation applied to this job went from console terminals (type in the code and see results directly), disk drives for storage, text editors ( vi vs. EMACS), interactive debugging and integrated development environment apps), higher level languages, open source software libraries, source version control like GIT, automatic syntax checking and error correction, automated testing and performance analysis and now moving to machine learning based “code-less”programming. This was all dependent on progress in semiconductors and other computer technology , human interface design. If someone tried to create modern software (say the app you are reading this on without the automation developed since 1976, it would be as difficult as it would be to build a 747 with only hand tools and paper drawings. I’ve worked at software for 50 years including at Apple and Microsoft and I can say that while the hours spent working never went down, the quality, amount and functionality of shipped software increased by orders of magnitude. Unfortunately my salary didn’t increase by the nearly same amount, but productivity did and I was able to capture some of that by investing in my employers’ stock. At no time did my job feel in danger of being lost to automation.
With github and python I can see programming moving closer and closer to natural language until at some point the technical skill required for fixing problems as they occur becomes something my six year old can figure out. I remember when I started and their was a big learning curve and you had to buy books on C++. Now you can just look at the library definitions or google for an answer. Some day soon this won't be required because every branch of the etl will be indexed with quality control systems that bring up the broken part so that a person of modest intellect doesn't have to search, can see why it broke and then try a few things to come up with a fix. Their are companies out their trying to make data searchable like a google search. Eventually that will become more like a siri. We aren't their yet but in 40 years we will be.
A very good review. I recall a few decades ago my area was populated with dozens of textile mills making everything a person might wear. It employed thousands of local workers doing all sorts of jobs from starting floor workers to management. They paid well, they offered benefits, they gave people a way of life. In less than a 10-year period these jobs were moved out of the U.S. to China and Mexico, and other countries where textile companies could produce for less. They prided themselves on helping developing countries develop.
Today with automation and AI these types of jobs can and should return to the U.S. to create more jobs at levels never before seen. Less floor workers and more advanced skills in mechanization, maintenance, computer integration, and management.
The robots are available, many from China. I say buy them up, even give tax or other incentives to buy them. Doing so could create a tsunami wave of new growth and move the U.S. back to a self manufactured country rather than a net buyer from everyone else. It's our loss of market direction in this area that fuels the countries who today would want to see us decline.
Heck yeah we need more robots!
Fascinating. It shows how difficult it is to predict trends. The exception to that might be demographics, and it would be interesting to see how robotics relates to that.
Really great article! There are lots of great reasons to bring manufacturing back (as you mentioned inflation reduction, supply chain bottlenecks, inequality) but I wanted your opinion on the risks associated with this as well. I imagine short term the benefits would outweigh but long term what about demand shifts to experiences rather than goods? Will millennials still prefer vacations to new cars long term and will that affect strategy for re-shoring manufacturing?
The issue is not robotic automation. It is robotic automation plus energy independence. The world is further along (perhaps much further along) on the path to accomplishing the first. When it accomplishes both, work will change completely.
Human labor can be divided into several components. Three primary ones are calculation, innovation, and mechanism. If they are given the same inputs and are asked to do the same calculation, our machines are much better than we are at calculation.
Innovation can be broadly thought of as ideation, pattern recognition, and error recognition. We are still doing better than the machines here. The goal of AI is to change that.
Much of the world's work, however, still involves the expenditure of human energy. Automation picks off those tasks as it becomes financially feasible to do so. Feasibility boils down to two things: Our ability to design replacement systems, and the cost of building them. We are getting better at the first every day. The second is a direct result of the cost of energy. Energy to power the machines that produce and modify the materials needed to build more machines.
Classical economic theory states that when a job is eliminated, the person holding that job moves on to what is now their highest and best use (i.e. the option that maximizes their utility). Two things need to be noted. First, this is not a guarantee that the new job will be better than the old one. In general, when all resources move to their highest use, society as a whole benefits. Individuals may not. Second, there is a floor below which it makes no sense for someone to accept their new roll. At this point, they will feel that they are better off going on the dole or turning to crime. Not everyone will be useful when most jobs involve programming.
It is hard to argue with any of the observations or conclusions in this article. The US certainly should not turn its back on any helpful technology, and being the best at important ones has got to be better than the alternative. But the best argument for using robots today is this: If you need to manage a tiger, it is best to start working with it when it is a cub. We need to consider what the world will look like in ten or 20 years.
> Not everyone will be useful when most jobs involve programming.
Programming will be automated long before nurse or occupational therapist or speech pathologist.
I think basically everyone will end up working in a human-human service role, for which the psychological requirements of the service basically by definition will exclude robots.
I am a software engineer of 10 years. Watching GitHub copilot’s progress in the last 14 months has me thinking my profession (as we know it, today) will be dead in 10 years.
Aren't economies dominated by human-to-human service roles historically notoriously terrible and exploitative ones reflective of a massive overproduction of humans relative to capital capable of productively employing them? Countries known for widespread use of domestic servants are, unsusprisingly, also known for attendant brutal labor exploitation and callous disregard for human rights. Being a domestic servant may be better than starving, but the widespread use and employment thereof seems like a much worse social equilibrium than the one we have now.
Obviously one rejoinder is that human to human service roles aren't limited to domestic servitude, but of course based on the remuneration to such roles in the present day -- whether daycare workers, home health aides, actors, or artists -- it would seem that we already have a relative glut of available labor even without massive displacement by capital, and that's notwithstanding the fact that at least daycare workers and home health aides are performing jobs with tremendous responsibility and value-add!
Naw. A surgeon just needs steady hands. He has a multi year degree for when things go wrong.
Yes, here surgeon is closer to software engineer than nurse. As you say, you’re not looking for human-human contact from a surgeon, just their steady hand.
But it the for when things go wrong that is the issue. That's when the human needs to be the expert.
"Our workers" can win against whom? Chinese workers? They're not the ones stealing from us, cheating us, usuring us, not paying us the value of our labor, keeping our wages low, our hours long, and our rents high. You're right that automation isn't the problem, but neither is China. The problem is capital.
Rentiers are often also capitalists, but it's the rent that's the problem. Georgist land value tax is the solution, and it could fund UBI which would give workers a much better bargaining position.
What's been described as "technological unemployment" is really just rentier capitalism at its core. The solution certainly isn't to smash up the robots and go full Hosni Mubarak on the Internet. Rather, it's to more fairly redistribute the robots and the Internet - including but not limited to breaking up Big Tech monopolies - so that they serve the public instead of dominating it.
It's very easy to forget that the original Luddites didn't sabotage weaving looms out of technophobia, they simply did so as a bargaining tactic for better pay and conditions, typically being the only ones who could get the looms moving again - a forerunner of the modern labour/trade union movement.