High unemployment rates, but even more efficient: AI and economics can only survive one.

Ask AI · If work is no longer necessary, how will human value be redefined?

AI Economics Series Part Two

#01

“Frenchification” of the UK’s economy

Last week I saw some UK economic data:

After the Labour government pushed for higher labor costs, the latest figures show the unemployment rate has risen to 5.2%, but at the same time business investment has grown by 3.5%, and labor productivity has increased by roughly 2%.

These data feel highly contradictory: on one side more people are unemployed, and on the other side businesses are more willing to invest and overall efficiency is higher.

Some economists summarize this shift in one sentence: the UK is “Frenchifying.”

In France, the characteristics of “high unemployment and high productivity” have existed for decades, due to features of the economic structure—such as a relatively high minimum wage, stronger labor protections, and higher social-security and hiring costs.

Under such a system, companies make a very rational choice: hire fewer people and use more machines. Behind this is a classic economic mechanism—capital substituting for labor.

Doesn’t this perfectly hit the “AI panic” many people have been feeling recently? The concerns we have in the AI era have actually been present in France for many years.

#02

A keyword that gets overlooked: capital deepening

In economics, this process is called “capital deepening,” which simply means: each employee is paired with more machines. Due to France’s unique historical culture and social-moral sense of form, the government provides strong protection for workers’ rights, and social welfare levels are relatively high—leading to expensive labor, making capital deepening a natural choice for firms.

That’s also why in France, labor productivity per hour in the long run can match—and even exceed—that of the United States among OECD countries.

This is a classic European paradox. It’s not that French people are truly more capable, but that “those who still have jobs” are highly efficient with the help of capital and machines, even with far fewer working hours than the U.S. In contrast, those who are not efficient enough have no work to do and are therefore not included in labor productivity data.

According to OECD global working-hours data, overall working hours in Europe are 15% fewer than in the United States—a situation that is the opposite of what it was before the 1970s.

About half of the gap can be attributed to vacation time. Europe typically has 4–6 weeks of vacation, including public holidays, while the U.S. has no federally mandated vacation requirements.

In addition, France and Belgium impose legal limits on the statutory workweek below 40 hours, plus a higher share of part-time jobs and work-sharing programs, along with limits on overtime pay. Meanwhile, the U.S. has maintained a relatively high work intensity among developed countries.

To maintain this labor structure, France’s industry mix is also distinctive: a large share of high-value-added sectors. For example, aerospace, luxury goods, nuclear power, pharmaceuticals, and high-end manufacturing are all capital-intensive + technology-intensive industries. They require only a small number of high-skill employees, along with a large amount of equipment and technical support.

Many European countries are similar to France. In fact, they have accepted this trade-off: higher productivity + more social protection + higher unemployment rates. Even though the UK has “left the EU,” in terms of economic structure, it has become even more “Europeanized” and “de-Americanized.”

Analyzing this characteristic of the French economy today is not about “being pro-France.” It’s that the development of AI may force more economies to “Frenchify” as well.

#03

AI: a more aggressive form of “capital substitution”

AI and automation are both typical technologies that substitute capital for labor. Companies will increase capital expenditures. This trend is already visible: data center investment is booming, and AI computing power investment is massive.

Correspondingly, employee hiring is declining. Although large-scale layoffs have not happened yet, that’s only because companies aren’t fully confident. They’re having employees learn how to use AI to work better—preparing them for the possibility that they’ll be replaced in the future.

In the discussion of whether AI will trigger a wave of unemployment, one side takes an optimistic view. They argue that over the past 200 years, each technological revolution replaced large numbers of jobs, but the long-term employment rate still rose, largely because new industries created more new job opportunities.

For example, the Industrial Revolution: artisans lost jobs, but more factory workers were created. In the electrification era, many artisan positions were eliminated, but new industries like automobiles and electrical products emerged. In the computer era, typists and telephone switchboard operators were eliminated, but new industries such as digital marketing and e-commerce were created.

But AI’s impact is completely different.

Automation uses industries that are labor-saving but intensive in low-end cognitive work to replace physical laborers. Since each employee’s productivity is roughly the same, it’s precisely this that allows more positions to be created and keeps the employment rate rising.

This time, AI begins by substituting for cognitive work. It can currently partly replace jobs such as customer service, translation, copywriting, programming, design, and data analysis—roles where efficiency is relatively not as high. But the jobs it can create right now are more productive. In purely quantitative terms, the new jobs AI creates in the future are very likely to be fewer than the jobs it “eliminates.”

If AI’s replacement speed exceeds the pace at which new occupations are created, then the “French-style” economic structure—**high productivity, high capital investment, and higher unemployment rates—**is likely to become a trend in the AI era.

Philosophers have discussed similar issues since the 20th century.

For example, Keynes proposed the concept of “technological unemployment,” and predicted that humanity might enter a society where people work only 15 hours per week; and Hannah Arendt went further, suggesting that if labor disappears, human beings would face a “crisis of meaning.”

So maybe it’s worth changing the perspective: if we don’t work, can we live well—maybe even better?

#04

Are we headed toward “full Frenchification”?

Longer-term lower employment rates in European countries have indeed brought a series of social problems. But European countries have gradually accepted high unemployment rates, focusing on solving the problems that come with them. The core logic is:

Even if you don’t work, you should still have a decent life.

This is the European welfare-state model, including high welfare protections, a comprehensive vocational training system, and a larger (relative to the U.S.) public sector that absorbs employment, plus shorter working hours to distribute work opportunities to more people.

However, in many European countries, this package hasn’t been very successful. The main reason is that when firms face competition from U.S. and East Asian companies, their productivity isn’t high enough to support high welfare.

But if AI significantly boosts overall social productivity, in theory it would be possible to have sufficient welfare, supporting a higher level of social income with less total social labor time.

That means humans no longer need “so much work.”

At the macro level that may be true, but at the individual level it’s different: “the same income, less work” can mean something completely different for different people:

1. For most people who treat work as a means of survival, AI can replace part of their jobs, but it cannot fully replace them. This is the first time in human history that people are liberated from labor.

2. For a subset of people who are completely replaced by AI but can’t easily find new employment, the traditional welfare system may no longer be enough. That’s when the “universal basic income (UBI)” plan—long discussed and debated repeatedly—should be activated: everyone receives a baseline income whether or not they work. I’ve analyzed this in depth in my article, “Directly giving money to residents might be a good way.”

3. For those who treat work as the meaning of life, this becomes a philosophical question: what is the meaning of work?

#05

A deeper question: the meaning of work

Let’s return to a more fundamental question: why do people work?

In the past, the answer was obvious. Work was both a way to survive and the basis of social class.

But in the AI era, **if production no longer depends on large amounts of human labor, then work may no longer be the core of how society runs—**and this is especially worth reflecting on for societies accustomed to “involution-style competition.”

When work is no longer closely tied to things like achievement, security, and identity at the social level, many people’s real problem becomes: if work is no longer necessary, how do we define our own value?

As the British writer Alain de Botton wrote in his book The Consolations of Philosophy, “Work might not bring happiness, but it keeps us from falling into a deeper kind of nihilism.”

The jobs that are newly created because of AI are made for people like this.