I think one reason for education being mentioned is that during 1999, in the midst of the 25 million layoffs, one policy was to rapidly expand higher education (sth like up 50% each year, I think) and help keep young ppl in college so laid off workers could look for jobs. Seems that keeping people in school longer (by encouraging ppl getting two master's, two PhD, or a PhD and then a master's) is always a strategy for job loss. But probably does not work now, as the grad exam enrollment is dropping. But in general, I guess education as a labor policy may work better in China, as ppl just generally see education as a good thing that you cannot have enough of.
Yea I’m curious how the general cultural inclination toward education will map onto the “终身学习” / lifelong learning idea. My (outsider) sense is that cultural inclination is very oriented toward formal academic education, something with structure and accreditation.
I think for the lifelong learning stuff to work in the context of AI, it’s going to have to be a much more fluid, constantly evolving and self-driven kind of ongoing “education”. Reform and opening showed that Chinese people can be incredibly adaptable, and I guess lots of that did involve this more fluid and self-driven form of education… yea even as I write this I’m questioning where I started from. Not sure where I land on this. Curious for any thoughts!
Think the 夜校经济 2 years ago could be an example of fluid life-long learning. Plus, there has always been 老年大学 at least in my city.
Also not sure if this lifelong learning is intended to re-train people to adapt to new roles AI creates, or shift all the meanings/prestige/social identities people used to assoc with work back to learning. Can also just be an excuse to extend formal education even longer until the workforce decreases as population declines...and all problems solved :)
My guess is the selected friction to layoffs would be temporary, once the AI industry generates more revenue, there will be more aggressive redistribution.
I do wonder if Xi’s “Do not turn into a welfare state” stays true now with the rapid AI development. I used to think UBI is unrealistic but now I’m starting to swing!
The Wuhan framing you describe is catching now partly because the humanoid manufacturing scale story is emerging at the same time. Leju Robotics just turned on the world's first automated humanoid production line in Foshan — 10,000 units per year, one every 30 minutes. JD.com launched robot maintenance/repair services in Beijing this week too, what they call 'robot care' — diagnostics, battery swaps for humanoids and AI companions. You don't build that service infrastructure for prototypes.
When that story breaks into mainstream Chinese media — and it will — does it read as industrial upgrading or as the next robotaxi moment?
I am not sure AI is posing the exact same problem for Chinese as to Americans though.
For one, a large portion of Chinese workers are still in physical labor. The kind of physical labor that isn't sitting in a car and largely remain static.
Then there is the "standardization" problem. The west, for all its good practice of applying equal treatment to whomever because of human rights, needs to comply and apply controlled framework to everybody, fairly. Those procedures are easy to be automated, because standards of outcome are clear.
It's hard to replace human for its creativity and agency in the west. But it is also hard to replace human for their relationship building, rules circumventing tendency in China. If a building needs to pass a fire resistance test with the fire department, the application on paper is the easy and cheap part to do in china. the one that really gets you through though, is how did you smooth out the relationship with the controller, the contractor and the client.
What China's AI anxiety reveals though, should be a warning if not a preview of our collective future: if you squeeze the population long enough and pushed them to the edge with no sense of security, they could collectively react in maniac and become even harder to control, and if they were give tools like AI/AGI, god knows how mindless they could push it.
Regarding Xi's negative views on the welfare state, these are not a departure from Marxist-Leninist thought but very much in the tradition of it. Rather than being linked to rugged individualist stances (as it often is in the West), it's more tied to the ideals of a worker's state.
Marx himself saw the welfare state as a capitalist sop. See also the concept of the Lumpenproletariat. While anti-welfare politics is coded as politically right in the West, it commonly exists amongst many ln the non-Western left as a productivist stance.
Interesting, thanks for sharing. I guess my question would be if you’re going to run a largely (state) capitalist economy where labor is often treated and compensated very poorly, is not providing welfare or a safety net really more Marxist? I can imagine an out-of-power revolutionary party being anti-welfare in order to heighten the contradictions(?) and accelerate the path to revolution. But if you’re the party in power not so much.
I think those are questions that will become more prominent when China's pie stops growing so rapidly. While it's true that China's labor conditions are poor in comparison to the West, they are not unusual compared to countries of similar GDP per capita. China's share of income to labour and gini index are better than Brazil, for example. I imagine Chinese economists judge their Marxist credentials on that kind of thing.
That said, China's work culture is more Stakhanovite than Marxist. Recent pieces in Chinese media on improving and enforcing time off indicates a rebalance may be underway.
We won't tax automated companies and distribute the accumulated wealth to the people, because that would be welfare. Instead, we will force the companies to keep around their human workers who no longer contribute anything to work essentially "bullshit" jobs. Because that's not welfare.
The key issue is not just that job anxiety is rising. It is that this changes the politically acceptable speed of deployment.
Once labor displacement becomes salient, the AI discussion in China stops being only about capability and competitiveness. It also becomes a sequencing question: how fast can deployment run before it starts colliding with social stability concerns?
#3 is very interesting and I would love to hear more of your thoughts on that. I don't see this point directly in the policy files, and my instinct is that the CCP should gesture toward protecting labor-intensive industries from AI rather than calling for more AI in these industries.
After all, stability/维稳 is one of the government's top mandates. But disrupting labor-intensive industries, which account for the majority of China's employment and where people have the weakest safety nets, goes against that mandate. On the other hand, the government has shown that it is willing to crack down on top earners in finance and tech. So the CCP might use a similar playbook for AI - frame AI as something that hurts the high earners while benefiting the rest.
The ‘don’t become a welfare state’ should be taken with a grain of salt given how many people are in unproductive but protected jobs at SOEs. Or have a look at the people managing gardens or people in subways. I.e. China might not want to pay for lying flat but has well before AI shown enormous willingness to provide subsidized jobs.
Great article, Matt. I didn’t realize Xi had spoken out so clearly against welfarism. It’s surprising to see the CCP facing slower growth for the first time in decades, while also dealing with worries about AI replacing jobs. For 45 years, the solution was always that the economy would keep expanding. What happens politically if that’s no longer the case, especially with AI making things move even faster? The labor arbitration ruling seems like an early attempt to respond, but it feels like no one has fully figured it out yet. Thanks for sharing this.
I think one reason for education being mentioned is that during 1999, in the midst of the 25 million layoffs, one policy was to rapidly expand higher education (sth like up 50% each year, I think) and help keep young ppl in college so laid off workers could look for jobs. Seems that keeping people in school longer (by encouraging ppl getting two master's, two PhD, or a PhD and then a master's) is always a strategy for job loss. But probably does not work now, as the grad exam enrollment is dropping. But in general, I guess education as a labor policy may work better in China, as ppl just generally see education as a good thing that you cannot have enough of.
Yea I’m curious how the general cultural inclination toward education will map onto the “终身学习” / lifelong learning idea. My (outsider) sense is that cultural inclination is very oriented toward formal academic education, something with structure and accreditation.
I think for the lifelong learning stuff to work in the context of AI, it’s going to have to be a much more fluid, constantly evolving and self-driven kind of ongoing “education”. Reform and opening showed that Chinese people can be incredibly adaptable, and I guess lots of that did involve this more fluid and self-driven form of education… yea even as I write this I’m questioning where I started from. Not sure where I land on this. Curious for any thoughts!
Think the 夜校经济 2 years ago could be an example of fluid life-long learning. Plus, there has always been 老年大学 at least in my city.
Also not sure if this lifelong learning is intended to re-train people to adapt to new roles AI creates, or shift all the meanings/prestige/social identities people used to assoc with work back to learning. Can also just be an excuse to extend formal education even longer until the workforce decreases as population declines...and all problems solved :)
My guess is the selected friction to layoffs would be temporary, once the AI industry generates more revenue, there will be more aggressive redistribution.
I do wonder if Xi’s “Do not turn into a welfare state” stays true now with the rapid AI development. I used to think UBI is unrealistic but now I’m starting to swing!
Walt Whitman?
The Wuhan framing you describe is catching now partly because the humanoid manufacturing scale story is emerging at the same time. Leju Robotics just turned on the world's first automated humanoid production line in Foshan — 10,000 units per year, one every 30 minutes. JD.com launched robot maintenance/repair services in Beijing this week too, what they call 'robot care' — diagnostics, battery swaps for humanoids and AI companions. You don't build that service infrastructure for prototypes.
When that story breaks into mainstream Chinese media — and it will — does it read as industrial upgrading or as the next robotaxi moment?
I am not sure AI is posing the exact same problem for Chinese as to Americans though.
For one, a large portion of Chinese workers are still in physical labor. The kind of physical labor that isn't sitting in a car and largely remain static.
Then there is the "standardization" problem. The west, for all its good practice of applying equal treatment to whomever because of human rights, needs to comply and apply controlled framework to everybody, fairly. Those procedures are easy to be automated, because standards of outcome are clear.
It's hard to replace human for its creativity and agency in the west. But it is also hard to replace human for their relationship building, rules circumventing tendency in China. If a building needs to pass a fire resistance test with the fire department, the application on paper is the easy and cheap part to do in china. the one that really gets you through though, is how did you smooth out the relationship with the controller, the contractor and the client.
What China's AI anxiety reveals though, should be a warning if not a preview of our collective future: if you squeeze the population long enough and pushed them to the edge with no sense of security, they could collectively react in maniac and become even harder to control, and if they were give tools like AI/AGI, god knows how mindless they could push it.
Regarding Xi's negative views on the welfare state, these are not a departure from Marxist-Leninist thought but very much in the tradition of it. Rather than being linked to rugged individualist stances (as it often is in the West), it's more tied to the ideals of a worker's state.
Marx himself saw the welfare state as a capitalist sop. See also the concept of the Lumpenproletariat. While anti-welfare politics is coded as politically right in the West, it commonly exists amongst many ln the non-Western left as a productivist stance.
Interesting, thanks for sharing. I guess my question would be if you’re going to run a largely (state) capitalist economy where labor is often treated and compensated very poorly, is not providing welfare or a safety net really more Marxist? I can imagine an out-of-power revolutionary party being anti-welfare in order to heighten the contradictions(?) and accelerate the path to revolution. But if you’re the party in power not so much.
I think those are questions that will become more prominent when China's pie stops growing so rapidly. While it's true that China's labor conditions are poor in comparison to the West, they are not unusual compared to countries of similar GDP per capita. China's share of income to labour and gini index are better than Brazil, for example. I imagine Chinese economists judge their Marxist credentials on that kind of thing.
That said, China's work culture is more Stakhanovite than Marxist. Recent pieces in Chinese media on improving and enforcing time off indicates a rebalance may be underway.
We won't tax automated companies and distribute the accumulated wealth to the people, because that would be welfare. Instead, we will force the companies to keep around their human workers who no longer contribute anything to work essentially "bullshit" jobs. Because that's not welfare.
The key issue is not just that job anxiety is rising. It is that this changes the politically acceptable speed of deployment.
Once labor displacement becomes salient, the AI discussion in China stops being only about capability and competitiveness. It also becomes a sequencing question: how fast can deployment run before it starts colliding with social stability concerns?
Terrific piece Matt
#3 is very interesting and I would love to hear more of your thoughts on that. I don't see this point directly in the policy files, and my instinct is that the CCP should gesture toward protecting labor-intensive industries from AI rather than calling for more AI in these industries.
After all, stability/维稳 is one of the government's top mandates. But disrupting labor-intensive industries, which account for the majority of China's employment and where people have the weakest safety nets, goes against that mandate. On the other hand, the government has shown that it is willing to crack down on top earners in finance and tech. So the CCP might use a similar playbook for AI - frame AI as something that hurts the high earners while benefiting the rest.
The ‘don’t become a welfare state’ should be taken with a grain of salt given how many people are in unproductive but protected jobs at SOEs. Or have a look at the people managing gardens or people in subways. I.e. China might not want to pay for lying flat but has well before AI shown enormous willingness to provide subsidized jobs.
It will be hard for AI to replace the subway security guards pretending to do security scans at the gates.
Great article, Matt. I didn’t realize Xi had spoken out so clearly against welfarism. It’s surprising to see the CCP facing slower growth for the first time in decades, while also dealing with worries about AI replacing jobs. For 45 years, the solution was always that the economy would keep expanding. What happens politically if that’s no longer the case, especially with AI making things move even faster? The labor arbitration ruling seems like an early attempt to respond, but it feels like no one has fully figured it out yet. Thanks for sharing this.