Unemployment, unbalanced distribution and structural changes: Can humans still "roll" over AI

Author: ginkgo

Source: Economic Observer

Image source: Generated by Unbounded AI tool

Human beings have long been concerned about AI, especially AI's impact on social employment and income distribution. Since the 1970s, we have experienced at least three waves of AI development. When the tide recedes round after round, people find that artificial intelligence does not seem to be as powerful as imagined, and they can't help but have reasons for more confidence and optimism. However, the development speed and capabilities of this round of AI seem to be different.

The emergence of ChatGPT (Generative Pre-Trained Transformer) and various generative AI tools enables humans to issue instructions to computers in natural language, which largely breaks certain professional barriers. Although the current AI-generated content still needs to be improved in terms of accuracy and originality, its ability to replace labor, reduce costs and increase efficiency is obvious. So, which occupations will this round of AI development impact, and will it bring a large number of new jobs as the optimists expect? In addition to trying to answer these two questions that have attracted much attention, the author also tries to analyze the structural changes in society brought about by AI, and what efforts individuals and society should make to deal with these changes.

We have seen that the current development of AI tools may lead to technical unemployment, deterioration of the income distribution structure, especially the "polarization" effect, and exacerbate various social problems. In order for technological progress to better realize inclusive value, we need to deeply reflect on the existing system and try to innovate and redesign the social system. In the final analysis, the social value realization and progress direction of technology are ultimately determined by human beings.

The Unemployment Crisis and the Possibility of Creating New Jobs

In the previous waves of automation technology, the impact was mainly on occupations that were repetitive or required less professional skills. However, AI large models have cognitive, analytical, reasoning, and creative abilities, so some occupations that are traditionally considered highly specialized may be greatly impacted.

A report by Challenger, a human resources consulting company, shows that in May 2023, the number of unemployed people caused by AI replacement in the United States will reach 3,900. This is the first time this report has cited AI as one of the main causes of unemployment, and all layoffs explicitly attributed to AI have occurred in the technology industry.

On the whole, I think occupations with the following characteristics may bear the brunt of the impact:

The first category is occupations that do not require high precision and standardization of results. For example, in art design, although AI is not yet able to generate top-level works, it can already produce "enough" designs. In related application scenarios such as poster design and marketing promotion material design, AIGC tools can already largely replace designers.

The second category consists of relatively auxiliary and entry-level positions within the knowledge and creative industries. For example, sorting out legal cases, relatively modular work in programming, etc. These jobs are often held by young people, so this kind of substitution will not only make it more difficult for young people to find employment, but also deprive them of the opportunity to continuously improve their professional capabilities at work. In the long run, it is extremely detrimental to the cultivation and accumulation of human resources in society as a whole.

The third category is jobs that are highly specialized but also highly modelable. This may surprise everyone the most. In fact, many jobs that require high professionalism have a high degree of modelability. For example, in the diagnosis process of medical workers, doctors used to learn a set of thinking and judgment models by studying a large number of cases, reading books and accumulating clinical experience. Now, this process can also be completed by AI, even faster and better.

The fourth category is industries with high individual needs but these needs have not been fully met at present. Such as education, escort and other industries. The impact these industries face may not be a simple substitution, but an industry reshuffle, that is, a transformation of industry entities and their organizational methods. For example, AI may promote the transformation of the form of education from one-to-many to more individualized teaching, and traditional institutions such as schools and hospitals may no longer dominate the industry.

Many people will point out that it will take some time for various substitution effects to emerge, so there is no need to worry about it now. Indeed, there are many factors that may slow down the large-scale deployment of AI, including the continuous development of application scenarios, social acceptance and recognition, supporting transformation of corresponding infrastructure, and the continuous establishment of laws and regulations. However, the improvement and evolution of AI capabilities are completely different from the speed at which carbon-based life forms learn and progress. If you look at the improvement in Midjourney's ability since it was released last year, you can intuitively feel the difference in speed. At present, the technical route based on the large model is relatively mature and clear, and the technical capabilities are constantly improving, coupled with the influx of capital and the formation of industry consensus, so the development and impact of AI will definitely come, and its speed is likely to exceed most people's expectations.

Some people also firmly believe that more new and better jobs will be created, as has been repeatedly confirmed in the previous process of human industrialization.

When we look at whether new jobs can be created, we can first look at how many new jobs are directly related to new technologies.

Just like before the advent of the mobile Internet era, we couldn't imagine the complex APP ecology and related jobs, so now we can't fully imagine the jobs related to AI. At present, it can be speculated that the continuous development of AI will bring about an increase in the demand for algorithm engineers, but this growth is limited, and the further development and prosperity of the digital world brought about by AI will be the source of more job growth.

For example, as the time and immersion of human beings living in the virtual world increase in the future, many product content design positions in the virtual world may be born. Avatar costumes, playbooks, game rules, etc. I think that with the rise of human demand for entertainment and spiritual life, the development prospects of creative industries are still considerable, and will continue to absorb and generate more jobs.

Second, we can look at the potential for new demand generation. This is firstly related to the characteristics of the times. For example, when humans entered the industrial society from an agricultural society, a large number of demands related to manufactured goods were not met, so the demand for related new products and services will be released in large quantities.

But in the age of AI, is there still such a huge potential demand space that has not been tapped? When people live in a more unified virtual world and digital space on a global scale, is there still a large new increase in diversified demand and supply? The author is not optimistic about this.

Be wary of deteriorating income distribution structure

Scholars have found that there is an important feature in the impact of the early automation process on income distribution, that is, the reduction of middle and low-skilled jobs and the growth of high-skilled jobs. According to Tuzemen and Willis' assessment, between 1983 and 2012, middle-skilled jobs fell from 59 percent to 45 percent, but high-skilled jobs rose from 26 percent to 37 percent. The overall change trend is relatively positive for social progress.

Although the application of large models is still in the very early stage, according to the research of Ali Zarifhona, a Ph.D. from Indiana University, it is predicted that some professionals and technicians will be most affected by AI in this round. In other words, the jobs that are most likely to be replaced are precisely those high-skilled, high-return jobs that are generally considered white-collar and gold-collar jobs. This group of people is precisely the group with the most spending power. The decline in their income levels and their poor expectations for job security will not only damage the income ratio of the middle and upper class, but also weaken the overall demand level and growth momentum of the economy. Its impact cannot be underestimated.

Everyone knows that a healthy social income distribution structure should be "spindle-shaped", thick in the middle and thin on both sides. However, the current development of AI tools will not only aggravate the imbalance between human labor and capital returns, but may also affect intelligence because of its impact on intelligence. The strengthening of the imbalance of returns further deepens the "polarization effect" of income distribution.

Because technology will change the importance of different factors of production in economic production, for example, the Industrial Revolution has significantly increased the importance of capital compared to human labor, and the large model has brought about a significant increase in the importance of intellectual factors. For example, an architectural designer may have been limited by his own efficiency in the past and needed to rely on a larger team and platform. But now if his ideas and reputation are good enough, he can receive more jobs than before through individuals or small teams. As a result, the most creative group of people will have higher intellectual added value and occupy more resources, resulting in the so-called "genius effect", occupying a higher and higher proportion of income distribution.

But this does not mean that the new technology is the "original sin", but requires people to adjust the income distribution system to avoid the continuous exacerbation of this extreme imbalance. The impact of technological progress should not be limited to the improvement of production efficiency, but also depends on the extent to which it can bring cross-industry spillover effects.

Regarding the cross-industry spillover effect, the most classic case comes from Ford. At the beginning of the last century, Ford improved the efficiency of automobile manufacturing by transforming the production process, but they did not stop there. Instead, they introduced supporting measures, which greatly reduced the price of automobiles, and at the same time greatly increased employee compensation, and also reduced the daily working hours from 9 hours reduced to 8 hours. Although the labor cost has increased significantly after the salary increase, the labor turnover rate has dropped significantly. High income supports workers to release various consumption demands, and they can also afford Ford-manufactured cars, which promotes the positive interaction and development between the supply side and the demand side. It became a textbook demonstration of cross-industry spillover effects.

Therefore, this growth model in which distribution and production promote each other is also called "Fordism" by some scholars. It tells us that in order for technological progress to better realize the value of universal benefits, it is also necessary to make changes to the existing institutional arrangements. adjustments, or even redesigns.

Because the 1% of the population at the top of the pyramid cannot generate the consumption power of 99% of the population, nor can it be converted into enough investment to absorb the existing human resources. An extremely unbalanced distribution structure will not only prevent cross-industry spillover effects from playing out, but will also exacerbate various social problems.

Liu Cixin told such an extreme story on a virtual planet in the short story "Caring for Humanity". In that world, the smartest person takes all the resources on the planet, while the others are driven to the earth, and the earth suffers as a result. We don't want this to happen, so we have to think ahead and plan ahead.

From individual to society, how to deal with structural changes

Regardless of whether our own jobs will be replaced by AI in the short term, when we gain insight into the structural changes AI brings to the social economy, we will find that in the near future, everyone may be affected by it.

The first change is that the degree of integration and systematization of the industry is getting higher and higher. As AI deepens into work and organization, it will not only change individual work, but the overall collaboration and organizational model of the industry. With the assistance of AI, more and more economic and social collaboration will enter the overall automation system, forming a huge and complex collaboration network between machines, and gradually excluding people. For example, if the automatic driving technology is fully implemented, the entire transportation system will become a whole, and the system command center will be like a brain, fully controlling the coordination and operation of various transportation equipment. This trend falls in every industry, which means that the integration potential of various industries has greatly increased. If the current industry concentration in your industry is relatively low, perhaps the current technological progress will bring huge innovation space, which is worth thinking about and grasping.

The second change is that the empowerment of top talents by AI will greatly enhance the industry competitiveness of small and medium-sized enterprises, and the resource advantages originally occupied by large enterprises will be further lost. Across all industries, we will observe a simultaneous intensification of the two trends of organizational miniaturization and industry concentration. Especially in some creative industries, such as advertising, games, architectural design, etc., this trend may be very obvious. In the field of AIGC itself, we can see the great energy of many small entrepreneurial companies. Midjouney, which subverted the entire design industry, is a company with only a dozen people.

The third is that the speed of change in the industry's competitive landscape and the speed of product iteration will further increase, and the "VUCA era" (VUCA) will become even more "Uka". The scope of this influence is extensive, and it is difficult for all walks of life to stay out of it, which requires entrepreneurs to have a clear understanding.

Faced with the above changes, as individuals, we are first required to adjust our focus of competence. Previous white-collar work focused on cognition and analysis. In the future, this will not be enough, at least general analysis will not be enough. We need to have stronger originality, insight, sensibility, exploration ability and creativity.

However, individual efforts cannot change the structural problems that AI will bring. When the development of technology has been integrated with human evolution, we must deeply reflect on the existing system and try to innovate and redesign the social system.

The first thing to reflect is our education system. Just imagine, if we have a child who loves painting, how should he/she deal with the powerful painting function of AIGC? Will they still be motivated to go through the long and painful practice process and finally become a painter who may not be the first-rate? However, if our children do not paint, read, or think now, but hand over these tasks to AI, how will they grow into a person with exploration and creativity? If we want to stimulate their creativity, where does their motivation come from? How can the education system provide these dynamics?

In addition, we need to change the current introversion society with a single standard. Efficiency has become a very important criterion for measuring individuals and organizations when human production capacity cannot meet human needs. But realistically speaking, in this dimension, carbon-based life will not be the opponent of silicon-based life after all. So, how should social development and progress be defined? In my personal opinion, there are two aspects that need to be taken into consideration at the same time.

On the one hand, we need to further increase the wealth of human society. But the key point here is how to define wealth. First of all, it needs to be emphasized that the core of wealth growth is not the accumulation of quantity, but the improvement of diversity. Why do we think that the wealth of human society now far exceeds any period in history? Not because there is more gold in the world, but because the richness of products and services that every ordinary individual can enjoy is unprecedented, and the accumulation of this diversity is the result of continuous human innovation. What needs to be made clear is that wealth is not just a material concept, and the definition of wealth in the future will be more and more emotional, cultural, spiritual and spiritual.

The measurement of the second aspect should lie in the development of human beings themselves. When technology shares the work of human beings, how can we enhance our inner sense of fullness and satisfaction? I think more and more individuals should have the opportunity to develop and pursue their own endogenous motivation in more diverse dimensions, and find their place in society.

In the final analysis, whether technology can be used by humans or replace humans does not depend on the technology itself, but on what kind of social system we want to build. If our institutions serve human greed, technology will reinforce this. But we can also use technology to better protect humans. For example, with the continuous development of production capacity, we can consider shortening the working hours per week, establishing a broader social security system, reducing the commercialization of people, and providing more protection for social and creative work, etc.

In fact, every leap in human production capacity does not naturally lead to an increase in the level of group welfare, and is often accompanied by a period of pain. Only by reflecting on and redesigning our institutional system, organizational goals, and value orientation, and every individual starts to think and act, can we drive society to evolve in a better direction.

(The author is Assistant Professor of Strategy and Entrepreneurship, China Europe International Business School)