How will artificial intelligence impact the future job market?

How will artificial intelligence impact the future job market?

More
请先 登录 后评论
  • 1 Follow
  • 0 Collect 50 Check
  • User 2023-12-12 13:23:33

1 + Answer

King of kings
擅长:Hacker,AI,Crack

Introduction

Artificial intelligence refers to a technical platform for computer applications to simulate human intelligence and achieve specific goals and tasks (Olsen&Tomlin, 2019; Kaplan&Haenlein, 2019; Acemoglu&Restrepo, 2020a). Since the Industrial Revolution opened the way for machines to replace human labor, discussions in the field of economics regarding the relationship between artificial intelligence and employment have been a controversial and unresolved topic for decades. The latest focus of discussion has shifted to the uniqueness of the new generation of artificial intelligence compared to past automation technologies, which further replaces mental labor and changes the division of labor, resulting in a differentiated value structure of human capital (Atack et al., 2019). With the widespread application of artificial intelligence in various industries, especially healthcare, agriculture, transportation, energy efficiency, and production environment control, significant benefits have been achieved (Plunkett, 2021). In recent years, some industrialized countries have faced the dilemma of declining labor force share and slow labor productivity growth while applying artificial intelligence on a large scale (Autor, 2019), which undoubtedly sends a clear negative signal to the labor market, Triggering a new round of debate in academia and industry over whether the new generation of artificial intelligence technology effectively promotes employment.

Existing research has conducted extensive discussions at different levels, from macro to micro, theoretical to empirical. One type of research employs large-scale quantitative analysis techniques, using multiple linear models and general equilibrium models to measure the likelihood of specific professions or even individual tasks being replaced by artificial intelligence, involving the United States, Germany, Finland, France China and other countries (Frey&Osborne, 2017; Arntz et al., 2016; Acemoglu&Restrepo, 2018a; Cheng Hong et al., 2018). Another type of research is based on technology task combination to analyze the creation and supplementation of artificial intelligence on employment at the micro level (Autor, 2015; Agrawal et al., 2019; Wang Zeyu, 2020). Scholars have also explored the impact path and mechanism of artificial intelligence on employment from a macro perspective, and concluded that artificial intelligence has different impacts on employment in the short and long term (Wang et al., 2017; Cheng Chengping, Peng Huan, 2018; He Qin et al., 2020). However, there are significant differences in the prediction results of the impact of artificial intelligence on the total employment and employment structure in the above studies, and there are limitations in the research paradigms and method models: the research is generally based on economic growth theory and unemployment theory, with a single research paradigm; Concentrate on analyzing the substitutive and creative effects of artificial intelligence on employment, and homogenize the research content; The economic model assumptions used have deviations from the actual situation, and the explanatory power of the research results is insufficient; The industry or job data used is not comprehensive and accurate enough. Although there have been comprehensive studies summarizing the development context, characteristics, current situation, and results of the impact of artificial intelligence on employment (Qiu Yue, He Qin, 2020; Long Yuntao et al., 2020), there is a lack of in-depth explanation and focused tracking of core content such as opposing views, contradictory conclusions, limitations, and unresolved issues in this field, which is not conducive to the deep promotion of research direction.

To promote the deep expansion of research on artificial intelligence and employment, this article conducts topic searches on the TOP5 economic journals, other SSCI indexed economic journals, research reports, and domestic CSSCI core journals through the Web of Science JCR core collection, JSTOR retroactive database, EBSCO database, OECD Economic Cooperation and Development Organization database, and China National Knowledge Infrastructure. The search terms include: Artistic Intelligence, Robot, Automation, Labor, Work, Job, Employment, artificial intelligence, robotics, labor, work, employment. Based on the content of the abstract, homogeneous literature was screened, resulting in 23 Chinese literature and 49 English literature, totaling 72 articles. Intended to discuss the following research questions: (1) What are the latest developments in the impact of artificial intelligence on total employment, employment structure, and mechanisms of action? What are the progress or limitations compared to past research results, and what are the reasons for these limitations? (2) Summarize the opposing or controversial viewpoints in this field. Are there any in-depth research questions that need to be explained? (3) Summarize the theoretical analysis paradigm and limitations of the impact of artificial intelligence on employment, and where should we expand in the future? The innovation of this article lies in focusing on the differences and new perspectives in the field of artificial intelligence and employment from a contradictory perspective, analyzing the reasons for these differences and new perspectives, and looking forward to the ideas and methods for resolving these differences. This forms a more in-depth and systematic knowledge system, which is conducive to exploring new research directions and providing a diversified research perspective for promoting employment stability and high-quality employment.

The main text is arranged as follows. Chapter 2: Sorting out the development context of industrial revolution and artificial intelligence, as well as the effects of artificial intelligence. Chapter 3: Sort out the opposing views and limitations of artificial intelligence on total employment, and explain the reasons for the limitations. Chapter 4 focuses on the concentrated manifestation of the impact of artificial intelligence on employment structure, analyzing the focus of debate, reasons, and influencing factors. Chapter 5: Sort out the mechanism of how artificial intelligence affects employment, extract the influencing factors, summarize the focus and limitations of each research stage, and propose new solutions. Chapter 6: Summarize the analysis results and research contributions, and propose future research prospects for unresolved issues.

Methodology of the Impact of Artificial Intelligence on Employment

Technology is considered to improve productivity and bring about labor savings (Phan et al., 2017). With the rapid development of artificial intelligence technology in the industrial revolution, the rearrangement of labor value creation and income distribution has led to increasingly intense discussions on the relationship between artificial intelligence and employment, as shown in Figure 1. Previous studies have found that the transition from manual production to machine production leads to significant labor substitution. Artificial intelligence replaces human labor by replacing cheaper capital (machines) in a series of tasks performed by humans, reducing the share of labor in value appreciation (Atack et al., 2019; Frank et al., 2019). The first two industrial revolutions brought about direct labor input savings through production mechanization and large-scale production, forming an intensive labor market (Liu Xiangli, 2020), transferring the share of agricultural labor to the manufacturing industry, increasing overall labor demand, and increasing demand for blue and white-collar jobs (Acemoglu&Restrepo, 2020a). The third industrial revolution used automation technology to enable labor and machines to simultaneously carry out work tasks, greatly increasing the skill intensity of labor. With the Fourth Industrial Revolution, artificial intelligence technology has developed rapidly, and advances in computer speed, data collection, data storage, and algorithms have led to a rapid increase in the level of artificial intelligence like human intelligence (Agrawal et al., 2019), which has had a more thorough impact on human employment. However, the contradiction between weak labor productivity growth and changes in job demand is deepening.

Figure 1: The timeline of the Industrial Revolution and the development of artificial intelligence

Image source: The author drew it themselves.

At present, the focus of debate is mainly on TechnoOptimists and Techno Pessimists. Technical pessimism believes that the phenomenon of technical unemployment needs to disappear on its own or rotate the impact of the next round of technological revolution, without a good solution (Autor et al., 2003). Technological optimism advocates increasing productivity to promote economic growth, creating new tasks to increase labor demand, and emphasizing the positive significance of artificial intelligence in increasing labor share (Acemoglu&Restrepo, 2020a). These two perspectives overlook the importance of artificial intelligence types. The adoption of the wrong type of artificial intelligence will result in markets, businesses, and individuals misallocating resources to labor-intensive tasks, exacerbating labor substitution, leading to severe unemployment, weak growth, and income inequality. Subsequently, many scholars have put forward different viewpoints. Fleming (2019) believes that there is a clear organizational boundary between the impact of artificial intelligence on employment. Technical unemployment is more dependent on the definition of whether work tasks can be automated by social organizational forces, and unemployment is more threatened by macroeconomic pressures on low skilled occupations. Pettersen (2019) continues to add that artificial intelligence does not pose a threat to knowledge-based work. Some scholars have proposed that artificial intelligence does not have an absolute favorable or harmful tendency towards employment, and a series of related economic factors need to be considered (Rotman, 2018). Therefore, it is necessary to further analyze the methodology of how artificial intelligence affects employment.

The academic research on methodology focuses on the field of microeconomics. The most representative research results come from the labor economics school led by professors Autor and Acemoglu from the Massachusetts Institute of Technology. They have been conducting a series of tracking model construction and empirical research on computerization, automation, artificial intelligence, and employment based on task models since the 1990s. Autor proposed a daily work task skill matching model (Autor et al., 2003; Acemoglu&Autor, 2011; Autor, 2015), which suggests that the degree to which workers are replaced by artificial intelligence depends on the matching degree between daily task types and skills. High skilled workers engaged in complex work tasks are not easily replaced by artificial intelligence due to their strong cognitive skills such as judgment, analysis, and problem-solving, while medium skilled workers engaged in procedural daily tasks are the most easily replaced, laying the theoretical foundation for task replacement tracking research. But Acemoglu and Restrepo (2018a) questioned the task substitution of highly skilled workers. Due to the increasing proximity of artificial intelligence technology to humans in decision-making, it has greatly increased the replacement rate of highly skilled and complex task workers in the US labor market. They further developed a comparative framework for the substitution of high and low skilled workers and found that high skilled workers exacerbated the ripple effect of low skilled workers being replaced during the substitution process. They believed that the root cause of excessive substitution was that when the wage rate was higher than the opportunity cost of labor, companies chose automation to save labor costs, while society considered the opportunity cost of labor and used automation less, The contradiction between the two leads to a significant substitution of medium skilled labor by artificial intelligence, which affects the long-term equilibrium of employment (Acemoglu&Restrepo, 2018b). The latest research confirms the above viewpoint that even workers who are highly exposed to artificial intelligence tasks have a negative impact on their employment and wage situation (Acemoglu et al., 2021).

The difference between the two types of methodologies lies in the different premise assumptions. Autor's research is based on the premise of isolated work tasks, which will expand the scope of tasks that capital can execute and the task substitution scope and substitution effect of automation on low skilled workers. And Acemoglu will lead the research direction towards the construction of continuous task models, incorporating technical type factors and task binding, breaking the advantage of high skilled workers over capital in complex tasks, indicating that capital not only has a relative advantage in low complexity daily manual tasks, but also has a relative advantage in complex tasks generated by high skilled labor, Emphasizing the relative comparative advantage of new complex tasks in promoting employment, namely the positive significance of productivity effects in maintaining employment stability and promoting labor force share growth, is a breakthrough in exploring comparative advantage models based on research on task skill combinations.

However, the above methodology also has limitations. (1) Sort tasks based on the relative advantage of labor and set them as continuous tasks, which is different from the actual order of executing tasks in production. (2) Neglecting the possibility of identical workers performing tasks before and after automation simultaneously. (3) Due to the combined influence of market and government factors (Cheng et al., 2019), the impact of artificial intelligence on employment is limited by a series of factors such as workplace, economic level, and education level (Clifton et al., 2020), making it difficult to obtain high-quality micro data. In the following research, overcoming these limitations requires improving the vertical and spatial resolution of data as well as workplace skill data. Overall, existing research has not yet specifically analyzed the modes of task transfer and labor division that artificial intelligence technology allows at different stages of development, and it is difficult to form a consistent judgment on the comprehensive impact of artificial intelligence on employment. There is limited research on the relationship between artificial intelligence and supply side driving factors such as work structure and job skills (Elliott, 2018), which fails to explain the uniqueness of the impact of artificial intelligence on employment compared to other technological changes. Therefore, some scholars have proposed a new breakthrough point (Lise&Vinay, 2020), which is to judge the potential of artificial intelligence to substantially replace employment from the perspective of human general skills. By improving general skills, the skill structure of future labor can achieve "matching and symbiosis" with artificial intelligence systems, and maximize the positive impact of artificial intelligence on the total employment and employment structure.

The Impact of Artificial Intelligence on Total Employment

The total employment reflects the actual utilization efficiency of labor resources during a certain period of time, and efficiency is usually intuitively reflected in changes in quantity. The first thing to consider is the increase or decrease in the total employment. Existing research has reached a consensus through theoretical models and empirical tests, confirming that artificial intelligence plays a role in the total employment through substitution effects, creation effects, and supplementation effects.

The impact of artificial intelligence on total employment is concentrated in two types of research. One type is theoretical research, which compares the size of various effects of artificial intelligence within the scope of capital and labor relations research, and uses the substitution elasticity between capital and labor to explain how artificial intelligence effects regulate short-term and medium-term total employment (Leon Ledesma&Satchi, 2019). (1) In the early stages of artificial intelligence applications, capital replacing labor triggers a displacement effect, also known as the displacement effect. Machine substitution of tasks originally performed by labor reduces labor demand, reduces the share of labor with added value, and causes a short-term decline in total employment (Acemoglu&Restrepo, 2019). It mainly replaces the programmatic and some non programmatic work that artificial intelligence excels at, such as handling, driving, and so on Imaging diagnosis, etc. (Autor, 2015; Liu Xiangli, 2020). (2) With the penetration of artificial intelligence among industries, capital creates labor through the manifestation of the Productivity Effect, also known as the productivity effect or spillover effect, which increases productivity and labor demand through flexible task allocation (Acemoglu&Restrepo, 2019). As artificial intelligence reduces costs and expands output, it also increases labor demand for non automated tasks and the production of other labor-intensive products, The creation effect will actually increase the total employment in the long run (Acemoglu&Restrepo, 2020b). (3) The creative effect can also generate a supplementary effect, also known as the restoration effect or reshaping effect, which is opposite to the substitution effect. By matching the capital skills of workers with technical skills, capital can supplement labor. Due to the substitution effect, the share of labor force in the industry

请先 登录 后评论