From Producing More to Distributing More: The Next Step for Social Development in the AI Era?

—Reconstructing Distribution Systems in the Age of Intelligence, as Seen Through the Lens of Public Employment Differences Between China, Spain, and the Nordics

Anyone who has lived in both China and Europe for an extended period often makes a similar observation:

Europe has been mired in an aging population, low birth rates, and slowing growth in traditional industries for years, yet the standard of living in most Western and Northern European countries remains impressive, and their economic vitality has not dissipated.

The truly worthwhile question behind this is not just about how good European welfare is, but a more fundamental one:

As a society’s productive efficiency continuously increases, where does the extra wealth ultimately flow?

A Counterintuitive Piece of Data

Many people have the impression that China has a large number of people “living off the public payroll”—civil servants, public institution employees, and so on.

But if we look at the proportion of public sector employees in the total employed population, the result might be the complete opposite of our intuition.

In 2023, the number of employees in China’s state organs and public institutions who made actual contributions to the housing provident fund was approximately 48.36 million, while the total national employment was about 740 million, according to the “National Housing Provident Fund Annual Report 2023”. Based on this data, employees in government agencies and public institutions account for about 6.5% of the total employed population. This scope includes government agencies and public institutions such as public schools and hospitals.

Now let’s look at Europe. According to the relatively standardized statistical scope of the OECD (“Government at a Glance 2025”):

Table

Spain’s proportion is more than double that of China, while Norway’s is four to five times higher.

This set of data shatters a common misconception: contrary to what many believe, the proportion of personnel directly allocated to government agencies and public institutions in China is not high relative to the total employed population.

Of course, the essence of the public sector is not just to “manage society,” but more importantly, to “serve society.” Education, healthcare, pensions, unemployment support, childcare, disability support—these are all public services.

So, what’s truly worth comparing is not who has more “civil servants managing people,” but rather:

How much of its productive output does a society use to guarantee and serve ordinary people.

Why Did Past Technological Revolutions Always Create New Jobs?

For hundreds of years, technological progress has continuously eliminated old jobs but has never triggered sustained, large-scale unemployment.

Agricultural mechanization displaced farm laborers, but factories opened their doors. Industrial automation replaced some assembly line workers, but commerce, finance, logistics, education, healthcare, and other service industries expanded to take their place.

Thus, we have formed a deep-seated belief:

Technology will eliminate some jobs, but it will always create more new ones.

This judgment held true in the past because machines of the past could only perform specific tasks.

A tractor could plow a field but couldn’t diagnose an illness; an industrial robot could weld but couldn’t understand language; a computer could calculate but couldn’t independently handle complex real-world environments.

Machines were responsible for “a specific task,” while humans retained the general abilities of understanding, judgment, communication, learning, and adaptation. Therefore, when one industry no longer needed as many people, workers could move into another industry that machines couldn’t handle.

But artificial intelligence and embodied intelligent robots are changing this premise.

AI is beginning to encroach on language, programming, design, analysis, customer service, administration, and professional knowledge work. Embodied intelligence, meanwhile, seeks to install understanding, judgment, and learning capabilities into robots, enabling them to recognize environments, understand commands, use tools, and complete multi-step tasks in the real world.

Studies have estimated that nearly 40% of global employment will be affected by AI, with this figure potentially reaching 60% in advanced economies. A significant portion of these jobs will see reduced labor demand as machines directly perform key tasks.

At the same time, general-purpose robot models have begun to attempt different physical operations based on language commands, extending AI’s reasoning capabilities from the digital world to the physical world.

This means that it’s not just factory workers who will be impacted in the future. Software developers—one of the professional groups most significantly affected by AI today—are seeing their work in code generation, debugging, testing, and documentation being rapidly restructured. The service industry, administrative work, transportation and delivery, business operations, and some professional technical positions will also be compressed simultaneously.

The old path of employment transition—“agricultural population moves to industry, industrial population moves to services”—may not be infinitely repeatable.

When AI can handle cognitive and reasoning tasks, and robots can handle physical tasks, it will be difficult for humans to find a new industry large enough, and perpetually inaccessible to machines, to absorb all the displaced workers.

Therefore, future policies can no longer be built on an unproven optimistic assumption: that when old jobs disappear, new ones will automatically appear in sufficient numbers to accommodate everyone.

Even if new occupations emerge in the future, they may not match the displaced jobs in terms of quantity, speed, or skill requirements.

This is the most fundamental difference between the age of artificial intelligence and past technological revolutions.

What Might Truly Break is the ‘Employment-Income-Consumption’ Chain

Modern economies have a hidden but crucial cycle:

Enterprises organize production → Residents participate in work → Residents earn wages → Use wages to buy goods and services → Enterprises earn revenue → Reproduce

This system works because production typically requires a large amount of human labor.

But when AI and robots can accomplish more and more production with fewer and fewer people, this chain may break.

An enterprise’s output can increase, and its profits can grow, but the number of employees may not increase in tandem. If machines replace a large number of jobs, and the profits generated by these machines primarily go to a few capital owners, an absurd situation arises:

More and more goods are produced.

Production costs get lower and lower.

Society’s material capacity becomes stronger and stronger.

But a large number of ordinary people, having lost their wage income, are unable to buy these goods.

Enterprises, facing insufficient demand, can only continue to cut prices, compress costs, and reduce employment. Residents, with unstable incomes, become even more reluctant to consume.

This forms another cycle:

Automation increases productivity → Labor demand decreases → Resident income and expectations come under pressure → Insufficient consumer demand → Intensified price competition among enterprises → Further decline in employment and income

At this point, what society truly lacks is not food, housing, cars, or various goods.

What’s lacking is: a distribution system capable of converting the productive output created by machines back into the income and purchasing power of ordinary people.

Spain and the Nordics: An Unfinished Bridge

Today, the high public employment rates in Spain and the Nordic countries already reflect an important institutional arrangement:

A portion of the wealth created by enterprises and workers is collected through taxes and public finance, and then transformed into education, healthcare, pensions, unemployment benefits, childcare, and other public services.

In this process, fiscal spending simultaneously produces two types of distribution outcomes:

On one hand, residents do not need to purchase basic healthcare, education, and social security entirely with their personal income.

On the other hand, public service personnel receive wages, which in turn become part of household income and social consumption.

Therefore, Norway’s nearly 30% public employment figure is not just a statistic of “how many people the government employs.” It shows that this society extracts a considerable portion of its productive output from the market income system and reinjects it into the entire society through public services and public employment.

Spain’s proportion is lower than that of the Nordics, but its public employment rate of around 15% is still significantly higher than China’s approximate rate. For people living in Spain, this system is not an abstract concept; it is embodied in public healthcare, schools, pensions, unemployment benefits, and local public services.

This system is not perfect, but it at least partially solves a problem: even if a person does not work in a high-profit industry, they are not completely deprived of healthcare, education, and basic living security.

In other words, it establishes a public distribution channel in addition to employment income.

Public Employment May Not Be the Final Answer for the Future

Today’s public welfare still requires a large number of people to provide services, which is why Nordic countries have high rates of public employment.

But as AI and embodied intelligence develop, public services themselves may also be automated. Robots can take on some nursing, cleaning, transport, monitoring, and rehabilitation tasks; AI can handle a large amount of administrative, medical auxiliary, teaching assistant, and social security review work.

If public services that once required a hundred people can be completed in the future by ten people working with machines, then there is no need to artificially preserve ninety jobs that are no longer necessary just to maintain the original public employment ratio.

Because the purpose of public services is to provide services to residents, not to keep people busy.

At this point, the distribution system must evolve further.

Today’s path is mainly:

Government receives revenue → Hires public service personnel → Staff receive wages → Residents receive public services

In the future, it may gradually become:

AI and robots create massive output → Society captures a portion of the automation gains → Through public services, social security, and direct income distribution → Allows all members of society to share in the fruits of production

In other words, today’s welfare states primarily achieve redistribution through “public employment”; the intelligent society of the future will need to gradually shift distribution from “earning income through work” to “sharing automation gains as a member of society.”

This is not about making everyone dependent on relief, but about acknowledging a new economic reality:

When society no longer needs everyone to work continuously, a basic living can no longer be entirely dependent on whether each person has a job.

Welfare is No Longer Just Relief, but a Fundamental System for Maintaining the Economic Cycle

Traditional views often see welfare as assistance for a minority of people in difficulty: temporary benefits for the unemployed, partial coverage for medical expenses, and pensions for the elderly.

But in a highly automated economy, the role of welfare may change fundamentally. It will no longer be just about helping a few who have fallen out of the labor market, but will take on a more universal function:

Converting the wealth created by machines into residents’ purchasing power.

Early capitalism witnessed absurd scenes: with a surplus of milk on the market, capitalists would rather pour it away than give it to the poor who couldn’t afford it. This was because under the system of the time, goods could only enter consumption through purchasing power; overproduction did not mean the poor could obtain these goods.

Today’s developed European countries have largely changed this situation through social welfare. Unemployment benefits, pensions, public healthcare, child allowances, and minimum income guarantees ensure that a person’s basic livelihood is no longer entirely dependent on their current market income.

The welfare system has not eliminated the market, but has added a channel outside the market for converting social wealth into residents’ basic consumption capacity.

If society can produce enough food, a person should not go hungry just because they are temporarily unemployed.

If the housing stock is ample, we cannot have large numbers of vacant properties while people in need of housing have nowhere to go.

If medical equipment and medicines are constantly improving, we cannot exclude some people from basic healthcare due to insufficient personal income.

Solving these problems does not mean the state directly allocates every single commodity. The market can still determine the production, price, and individual choice for a vast range of goods.

What the public system needs to guarantee is: that every individual has the actual ability to access basic living materials and public services.

This ability can come from public healthcare, public education, pensions, and unemployment benefits, or it could come from housing subsidies, child allowances, minimum income guarantees, negative income tax, social dividends, or some form of basic income in the future.

The specific method is open for discussion, but the key principle is one:

We cannot let the increase in productive capacity ultimately manifest only as a surplus of goods and a lack of purchasing power among ordinary people.

China Needs to Prepare Its Systems for an Era of ‘Not Enough Jobs’

For the past few decades, China’s development has focused on expanding production, construction, and employment. This was entirely necessary during the periods of industrialization and urbanization.

But the real estate sector has stopped expanding, roads and industrial parks cannot be built indefinitely, and the population cannot grow forever.

At the same time, China has one of the world’s largest manufacturing systems and is rapidly developing AI, industrial robots, and automation equipment.

This means China is likely to encounter a new contradiction sooner than many other countries: powerful production capacity coexisting with insufficient domestic demand.

In this situation, simply demanding that companies continue to create more jobs or requiring the unemployed to constantly learn new skills may not be enough. Training is still meaningful, and new occupations will still emerge, but policy must simultaneously prepare for another possibility:

Even if everyone is trained, society may no longer need that much human labor.

Therefore, the next phase of China’s institutional design should gradually expand from “how to create a job for everyone” to “how to ensure everyone still has a basic income, public services, and a dignified life, even when society no longer needs so many jobs.”

First, establish a basic safety net that is not entirely dependent on employment status.

Today, many social security benefits are still tightly bound to one’s employer, years of contribution, and stable employment. This system is suitable for an industrial society centered on long-term employment, but it may not be suitable for a future of frequent job changes, an increase in freelance work, and large-scale automation.

Healthcare, pensions, child support, unemployment support, and minimum living guarantees need to become more universal and portable. A person who loses their job should not simultaneously lose their health coverage, housing security, and basic means of subsistence.

Social security must gradually shift from being “built around the employer” to being “built around the person.”

Second, establish a social sharing mechanism for the gains from automation.

When AI and robots increase productivity, the benefits will first be reflected in corporate profits, capital value, and the revenues of a few key platforms. If there is no new connection between these gains and residents’ incomes, the wealth gap and insufficient demand could widen simultaneously.

We can gradually explore:

• Steadily using more capital gains for social security.
• Expanding long-term capital sources for social security funds.
• Establishing a reasonable tax and distribution mechanism for the excess profits brought by automation.
• Allowing public capital to participate in key AI, computing power, and infrastructure construction in appropriate ways, with the long-term returns shared by society.
• In regions where conditions are ripe, exploring pilot programs for social dividends, negative income tax, or basic income-like systems.

The point is not to simply tax every robot—taxing technology itself may inhibit efficiency gains. The real problem to solve is: ensuring a stable portion of the gains from automation can flow into the income and security systems of all residents.

Third, shift more fiscal resources from “expanding construction” to “securing residents’ well-being.”

In the past, real estate, land development, and infrastructure investment played important roles in creating demand, absorbing employment, and supporting local government finances. But as the real estate sector enters a period of adjustment, this cycle is becoming increasingly difficult to sustain.

A more stable domestic demand cycle in the future might be:

Technology increases productivity → Corporate and social wealth increases → Resident income and social security improve → Families’ worries about future risks decrease → Consumption capacity and willingness increase → Enterprises secure stable demand

Healthcare, pensions, child subsidies, unemployment benefits, and housing support are not just social welfare; they are also the economic infrastructure that transforms productive capacity into effective demand.

Shifting fiscal spending from “primarily investing in things” to “primarily securing people” does not mean abandoning development. It means that after productive capacity is already strong enough, we need to find the real users for the fruits of that production.

Fourth, have the public sector increase services, not management.

China’s low public employment ratio does not mean we should simply expand all government agencies. What the public sector needs to strengthen is its service capacity that ordinary people can directly feel, not repetitive approvals, layers of inspections, and unnecessary administrative management.

Public resources should be directed more towards:

• Primary healthcare and long-term care
• Elderly and disability services
• Childcare and educational support
• Income support and transition services for the unemployed
• Basic housing and community public services

The standard for evaluating the public sector should not just be how many jobs were added, but whether:

• Residents have truly received services?
• The burdens of healthcare and education have decreased?
• The unemployed are able to maintain a basic living?
• The elderly and vulnerable groups are being supported?
• Fiscal investment has been converted into residents’ sense of security and actual purchasing power?

The fundamental purpose of the public sector is to serve people, not manage them.

Fifth, build “automatic stabilizers” for the distribution system.

When the economy slows, unemployment rises, and incomes fall, if all support requires case-by-case applications and temporary approvals, policy often arrives too late.

In the future, we can build stronger automatic adjustment mechanisms: when the unemployment rate rises or resident income falls significantly, unemployment benefits, minimum income support, housing subsidies, and child allowances can automatically expand; when the economy recovers and incomes rise, some temporary support can be gradually phased out.

This can both prevent mass unemployment from quickly turning into a consumption collapse and reduce reliance on temporary projects and administrative relief during every economic downturn.

Social security is not an optional extra to be considered only after economic growth is achieved; it is itself a part of maintaining the stability of the economic cycle.

What Really Needs to Change is How We Frame the Problem of Development

In the past, we asked:

How to produce more food?

How to build more housing?

How to establish more factories?

How to create more jobs?

These questions were once very important, and they have not completely lost their meaning today.

But in the age of artificial intelligence and robots, we must add a new question:

When machines can perform more and more production and services, how does society distribute these achievements to people fairly?

This is not about opposing technological progress. On the contrary, only by ensuring that the fruits of increased productivity are more widely shared can technological progress be sustainable.

Otherwise, the most absurd outcome could occur:

• Machines can produce everything, but companies cannot find enough consumers.
• Society has abundant goods, but ordinary people cannot obtain them because they have no wages.
• Productive efficiency reaches unprecedented heights, yet people’s lives become more unstable.

What We Lack is a Second Bridge Between Productivity and Ordinary People

Industrial society primarily relies on wages to connect the fruits of production with ordinary people. People participate in labor, earn wages, and then use those wages to buy products.

Wages are the first bridge between production and consumption.

But if AI and robots continuously erode human employment, this bridge may become increasingly narrow.

In the future, a second bridge must be built:

Public services, social security, returns from public capital, and social dividends, together connecting the wealth created by automation to the lives of ordinary people.

Spain and the Nordic countries’ high public employment rates today can be seen as an early form of this bridge. Through public finance, they convert a portion of social wealth into public services, resident security, and public sector income.

In the future, when robots can even take on a large number of public services, this bridge should not disappear. Instead, it should become more direct—distribution no longer needs to be mediated by the creation of a job.

The truly important policy in the age of artificial intelligence is not to prevent machines from replacing people, nor to forcibly create a batch of jobs that machines could also do.

It is to ensure that:

The prosperity created by machines does not come at the cost of the majority losing their income, purchasing power, and dignity.

When AI and robots make our productive capacity powerful enough, what we truly lack may no longer be more goods, nor more jobs created just to maintain employment.

What we truly lack is a distribution system that matches this new productive capacity:

One that allows the food that can be produced to reach the hands of those who need to eat;

One that allows existing housing to serve real housing needs;

One that allows medical capabilities to reach patients;

One that allows the wealth created by machines to ultimately be transformed into a secure life for all of society.

Only by completing this step will the productivity explosion brought by artificial intelligence become a true advancement for all of society, not just a capital feast for a few.