Liu Cheng/Wen

In the early stages of the Industrial Revolution in Britain, German philosopher and revolutionary Friedrich Engels observed a phenomenon: although machine based industrialization greatly improved social productivity, the living conditions of the working class continued to deteriorate for a considerable period of time, and the benefits of technological progress did not immediately benefit ordinary workers.

This historic 'pause' can be summarized as a time lag effect between technological revolution and human well-being. Nowadays, as the wave of artificial intelligence sweeps across the globe, a similar question is emerging: is the AI era also experiencing some kind of 'Engels pause'?

The two significant reversals in people's understanding of the intelligent economy that have sparked this attention are: firstly, there were doubts in society about whether AI could truly promote economic growth, and the shadow of the Solow paradox lingered. However, now this doubt has basically dissipated, and the driving role of AI in productivity and industrial transition is becoming increasingly clear; Secondly, in the past, people generally believed that the impact of AI on employment was still an illusion and had not manifested on a large scale in reality. However, since the beginning of this year, from the short drama industry to program development, from ride hailing to logistics and distribution, the AI substitution effect has suddenly exploded at a visible speed in multiple fields. The two reversals overlap, and the signs of "growth without income" pointed out by Engels' pause are now faintly visible.

The role of AI in economic growth and industrial transition

Looking back at history, every major technological revolution has been accompanied by a similar cycle of questioning.

In the Internet era, people have long debated whether information technology really improves productivity; In the early stages of the digital wave, the "Solow paradox" - that computers are everywhere but cannot be seen in productivity statistics - became a hot topic in academia.

Economists have proposed multiple explanations for this: it takes a long time from technology research and development, equipment installation to application scenario implementation; The full release of new technologies requires significant investment in complementary infrastructure; The supporting measures for enterprise organizational form, business process reengineering, and social system reform are seriously lagging behind; A large number of new business formats and services have not been effectively covered by traditional statistical systems.

Whether it is networking, informatization, or digitization, over time, people have generally accepted the fact that these new technologies can significantly improve production efficiency. Nowadays, the AI era is also facing the questioning of the Solow paradox, and it is worth noting that the duration of this questioning has significantly shortened.

Unlike the paradoxical period of more than ten years in previous technological revolutions, the cycle of AI from sparking controversy to demonstrating actual economic benefits has been significantly compressed. People quickly see the substantial driving role of AI in economic growth, total factor productivity, and industrial transition.

Since 2026, the driving effect of China's AI related industries on the macro economy has fully emerged.

From the investment side, AI infrastructure investment has continued to grow at a high speed, and new infrastructure such as computing center, Token factory, and large model training cluster have become the focus of competitive layout, driving the growth of fixed assets investment back up.

From the perspective of the industry, the profits of industrial enterprises empowered by AI have improved, and application scenarios such as intelligent manufacturing, intelligent quality inspection, and intelligent scheduling have been widely implemented, promoting a high growth rate of industrial added value.

From the perspective of trade, the export growth of AI driven intelligent products is strong, and the competitiveness of high value-added products such as intelligent driving cars, AI terminal equipment, and intelligent robots in the international market continues to increase.

It is worth noting that the consumption of large model keywords is exploding, and the demand for computing power continues to rise. This not only reflects the rapid expansion of the depth and breadth of AI applications, but also means that a huge intelligent economic ecosystem is accelerating its formation.

Not only in macro statistical data, ordinary people can also personally feel the penetration of AI.

AI products have become ubiquitous in various scenarios of work and life: robotic arms, remote-controlled robots entering factory workshops, Spring Festival Gala stages, exhibition halls, sports fields, stock markets, commercial operation of autonomous taxis in multiple cities, wearable devices such as AI glasses moving from concept to mass production, AI comics entering smartphone screens, TV screens, and movie screens, reshaping the landscape of the content industry.

Taking AI short dramas as an example, by 2026, over 95% of micro dramas have introduced AI generation technology in the production process, covering almost the entire process from script creation, storyboard design to image generation, dubbing and music. While the production cost has been significantly reduced, the output efficiency has been improved several times. The explosive growth in this field is an extreme epitome of the empowering effect of the AI industry.

From a deeper perspective, the driving force of AI on economic growth is not simply tool substitution, but is triggering a fundamental leap in industrial paradigms, which is a positive manifestation of industrial structure upgrading and accelerated transformation of old and new driving forces.

The path of upgrading traditional industries is linear evolution, while the changes brought by AI are non-linear, cross disciplinary, and systematic. It not only improves efficiency in a single link, but also reconstructs the value distribution logic of the entire industry chain - from research and development design to production and manufacturing, from marketing services to after-sales operation and maintenance, AI is creating new possibilities at every node.

This systematic industrial transition is the core characteristic that distinguishes the AI era from previous technological revolutions, and it is also the fundamental reason why the Solow paradox was rapidly broken in the AI era.

The substantial impact of AI on employment and income

However, the other side of the coin is also clearly visible. AI is driving economic growth while also having unprecedented substantive impacts on the job market and income distribution.

The author proposed in the first half of 2025 that "new digital economy jobs may become the hardest hit areas for layoffs in the AI era", which has now been fully verified by reality.

Taking the short drama live streaming industry as an example, this field, which was considered a new blue ocean in the digital economy two years ago, is undergoing a drastic adjustment under the impact of AI. With the maturity of AI generation technology, a large number of jobs that originally required manual completion such as screenwriting, editing, dubbing, and special effects have been replaced by AI, resulting in a significant decline in income for practitioners and a sharp reduction in the number of positions.

Hengdian Film and Television City, once a barometer of China's film and television industry, has experienced a wave of mass actor unemployment, with a large number of temporary actors and extras losing their livelihood job opportunities. The root cause of this phenomenon is the comprehensive penetration of AI into the production process of film and television content.

What is even more alarming is that the tentacles of AI substitution are extending from low skilled positions to high skilled mental labor fields. Programmers - a profession once considered "irreplaceable" in the AI era - are experiencing unprecedented impact.

With the rapid iteration of AI programming assistants, a large amount of basic code writing, debugging, and testing work has been efficiently completed by AI, greatly compressing the employment space for programmers. Some Internet enterprises have begun to reduce the size of their technical teams, using fewer senior engineers and AI tools to complete projects that used to require a lot of human investment. The breadth and depth of this substitution far exceed the general expectations of society before.

In the fields of physical labor and service industry, the substitution effect is also accelerating.

Ride hailing drivers and food delivery drivers - these two "reservoir" industries that absorb a large amount of labor - are facing direct competition from autonomous driving and intelligent delivery systems.

The Ministry of Industry and Information Technology and other departments have released the "15th Five Year Plan for the Development of Intelligent Connected New Energy Vehicle Industry", which proposes to achieve highly autonomous driving in scenarios such as highways, urban expressways, and some urban roads by 2030. This means that the employment substitution in the ride hailing and delivery industries will shift from a "potential risk" to a "real process".

At the same time, some platform companies have begun to collect and distill work data from frontline workers such as delivery personnel and household workers on a large scale. Once these data are fully learned and internalized, they will be directly transformed into the capabilities of automated systems, further accelerating the replacement of manual labor.

The profoundness of this substitution lies in its fundamental difference from previous technological revolutions: AI's substitution of labor has a wider scope and greater intensity.

In the era of cars replacing horse drawn carriages, although carriage drivers lost their jobs, they could transform into car drivers, and new technologies created new and larger scale job opportunities.

When autonomous driving replaces manual driving, humans are completely "abandoned": the number of backend controllers and maintenance personnel is extremely small compared to the original driver group, making it impossible to absorb the replaced labor force. The number of new jobs brought by AI is far from enough to compensate for the eliminated old jobs, and this "net reduction" effect is unprecedented.

In the United States, although the overall employment rate remains high and the unemployment caused by AI is not yet apparent in macro data, multiple studies have begun to pay high attention to the issue of job substitution, and multiple surveys have shown that AI related job losses are accelerating and accumulating.

In 2025 and early this year, people are still paying more attention to the empowerment of AI. At that time, the empowering effect of AI on individuals was more significant: AI was a tool, and people used AI to improve efficiency.

Nowadays, AI has shifted from "empowering" to "replacing" in certain industries, from an auxiliary role to a protagonist, and the speed of its role change is unprecedented. 95% of micro dramas are generated by AI, and Hengdian Group plays the unemployment wave. These are no longer isolated cases, but a microcosm of the AI employment replacement wave.

When AI can not only assist in creation, but also independently complete the entire process from planning to film production, a large number of intermediate practitioners lose their value.

This transition from "human-machine collaboration" to "machine solo" is the most vivid manifestation of the "Engels pause" in the AI era: technology has advanced, but the situation of a considerable number of workers is deteriorating.

Policy Implications

In the face of this trend, we first need to recognize that the new round of technological revolution and industrial transformation driven by AI is irreversible. Under the multiple driving forces of intensified international competition, companies pursuing profit maximization, and consumers continuously seeking new products and experiences, this process will not stagnate due to the interests and demands of any single group. Everyone can only face it calmly, actively improve their skills, and adapt to the new paradigm of human-machine collaboration.

For the whole society, as a general-purpose technology, AI generally has more advantages than disadvantages: it improves production efficiency, creates new products and services, and promotes economic growth.

However, locally affected groups may fall into a 'prisoner's dilemma': they will not (and will find it difficult) to consciously resist technological progress, and each individual cannot change the trend alone.

The helpless situation in reality is that while they cooperate with the enterprise to collect and distill their own work data, they clearly know that they may soon be replaced by the system trained from these data; While enjoying the convenience of life and work brought by AI, they have to adapt to the trend of skill evolution and reposition themselves.

The contradiction between individual rationality and collective dilemma is the core challenge facing social governance in the AI era.

Based on the above analysis, the author proposes the following policy recommendations.

Firstly, strengthen employment protection for vulnerable groups and establish specialized employment guarantee mechanisms for industries and groups where AI substitution effects are most concentrated. For high-risk groups such as short drama live streaming practitioners, ride hailing drivers, and delivery personnel, we provide job transfer training, skill enhancement, and transitional living security. Explore the establishment of "AI replacement insurance" or "technical unemployment fund" to provide a safety net for workers who lose their jobs due to AI replacement.

Secondly, retain necessary employment opportunities for workers, while promoting the application of AI, and reserve a certain proportion of manual positions in areas related to public safety, humanistic care, complex decision-making, etc. In the field of public services, avoid the dehumanization of services caused by excessive automation. Encourage enterprises to seek a balance between efficiency and employment through policy guidance.

Thirdly, increasing the initial share of labor distribution, the productivity improvement brought by AI should not only be translated into enterprise profits and capital returns, but workers should also share the dividends of technological progress. Explore the establishment of a sharing mechanism for AI productivity, through wage negotiations, profit sharing, equity incentives, and other means, to ensure that workers' income levels do not decrease but instead increase in the AI era.

Fourthly, we need to increase the government's ability and efforts in redistributing income. Faced with the potential exacerbation of income inequality by AI, the government needs to strengthen its redistributive functions. Explore levying "automation tax" or "robot tax" on enterprises that replace labor with AI, and use the proceeds for investment in social security and public services. At the same time, research and formulate institutional designs such as universal basic income and negative income tax to ensure that the social security network in the AI era is sufficiently solid.

Engels' pause is not an insurmountable historical fate. During the Industrial Revolution, after decades of institutional adjustments and social changes, technological progress ultimately benefited a wider range of groups. In the era of AI, we also have the opportunity to shorten the pause period through forward-looking policy design and institutional innovation, so that the dividends of AI can benefit everyone faster and more fairly. The key is whether we can always put human well-being at the center of development while technology advances rapidly.

(The author is a researcher at the Institute of Finance and Economics Strategy, Chinese Academy of Social Sciences)

(Author Liu Cheng)