Artificial intelligence and labor market outcomes

Elevator pitch

Artificial intelligence (AI) has streamlined processes, improved workforce allocation, and created new jobs to meet the needs of digitalization and automation. Individuals with AI capital experience greater employment opportunities and higher wages, particularly in high-skilled roles and large firms. Training in AI helps reduce gender-based digital disparities, empowers individuals, and enhances their employability. Policymakers should promote inclusive AI development policies to prevent widening AI-related divides and unemployment, and to ensure equitable opportunities.

Key findings

Author's main message

AI is reshaping the labor market by creating new jobs and increasing competition for high-skilled roles, benefiting those with AI capital. While AI may boost productivity in certain jobs, it also widens the gap between high- and low-skilled employees. Less-educated employees face higher risks of displacement and reduced income. Additionally, AI introduces challenges related to workforce adaptability, trust, ethics, and transparency, which negatively impact employees' job realities. Policymakers should navigate these changes to maximize the benefits of AI while mitigating its adverse effects.

Motivation

Governments are proactively working to maximize the benefits of artificial intelligence (AI) in their economies and labor markets. In 2021, the UK government released the National AI Strategy, aiming to boost business use of AI, attract international investment, and develop the next generation of technology talent. Similarly, in 2019, the US launched the American AI Initiative with the mission to promote leadership in AI research, development, and application, as well as to expand education and training opportunities to prepare the American workforce for the new era of AI. Examining the association between AI and labor outcomes is critically important due to its profound and multifaceted impact at both the macro level—affecting labor markets and economic dynamics across different regions—and the micro level, influencing individual and firm behaviors [1], [2], [3], [4], [5], [6], [7], [8].

Discussion of pros and cons

Macro- and micro-level evidence on the relationship between AI and labor market outcomes

The rapid advancements in AI technologies and systems are reshaping industries, influencing operations and job structures, and altering the demand for various skills, including digital literacy, data management, model training, decision-making, creativity, and innovation [1], [2]. In both the US and Europe, AI has the potential to expose up to two-thirds of current jobs to some degree of automation [9]. Advanced economies are likely to experience more significant effects from AI automation than emerging markets, due to differences in labor market composition and industrial structures [9], [10]. In the UK in 2020, at least 42.5% of small and medium-sized enterprises (SMEs) utilized AI in their operations [2]. As observed in Figure 1, UK SMEs used AI for various operations, such as collecting information related to customers' online purchase histories and digital footprints, offering cash flow forecasting, protecting data, strengthening cybersecurity, and providing legal and HR services [2].

International macro-level evidence highlights the importance of examining how AI is associated with labor market outcomes, with the magnitude of AI’s impact varying across studies. For instance, in the US, between 2010 and 2018, a one-standard-deviation increase in investments in education and skills related to AI led to a 19.5% increase in sales, an 18.1% increase in employment, and a 22.3% increase in market valuation [3]. Additionally, between 2010 and 2019, there was a notable rise in the demand for AI skills across US industries, accompanied by significant wage premiums, particularly in high-skilled occupations [7]. On average, vacancies that required AI skills offered an 11% higher salary within the same firm and a 5% premium within the same job title compared to positions that did not require AI skills [7].

In China, evidence from 30 provinces between 2006 and 2020 indicates that AI adoption, particularly through industrial robots, has positively impacted employment [4]. Notably, the scale effect resulting from advanced productivity leads to lower product prices and higher employee incomes, which in turn drives increased demand and economic growth, ultimately leading to greater output growth and employment [4]. Expanding the geographic focus, evidence from 23 OECD countries between 2012 and 2019 shows a positive association between AI exposure and employment growth in occupations with high computer use, suggesting that AI can complement human labor in high-skilled sectors [6]. 

These patterns highlight AI's potential to drive employment and growth by enabling more efficient production and fostering new industries [9]. Indeed, predictions suggest that AI has the potential to increase global GDP growth by 7% and by 1.5% in the US, demonstrating substantial macroeconomic potential if AI adoption becomes widespread [9]. More modest predictions in the US indicate that AI is projected to increase total factor productivity by no more than 0.66% [11].

At a micro level, in the UK in 2022, job applicants with AI capital experienced approximately 14.4% higher wage prospects compared to those without AI capital [1]. The illustration shows that in business administration occupations, the wage premium was the highest in the sample, followed by accounting occupations and economics occupations.

Similarly, as observed in Figure 2, job applicants with AI capital had, on average, a 22.3 percentage point higher chance of receiving interview invitations than those without AI capital [1]. In economics, entrepreneurship, business administration, and finance occupations, the percentage point change difference in interview invitations between those with and without AI capital are the highest. These findings highlight the need for education, job retraining, and innovation support to ensure a fair and inclusive transition to an AI-driven economy  [1].

At the same time, macro-level evidence in the US has found that firms with higher initial shares of highly educated employees and STEM employees invest more in AI [12]. As these firms invest in AI, they tend to transition to more educated workforces, with higher shares of employees holding undergraduate and graduate degrees, and greater specialization in STEM fields and IT skills [12]. Additionally, AI adoption in the US has been associated with reduced hiring for non-AI positions, indicating a potential substitution effect where AI replaces tasks traditionally performed by humans [5]. 

Given the nature of the phenomenon under examination, the following sections attempt to group and present the associations between AI and advancements in labor market outcomes, as well as the associations between AI and drawbacks in the labor market.

AI and advancements in labor market outcomes

Literature reviews have found that AI is transforming the labor market by creating new jobs that did not previously exist, particularly to meet the demands of digitalization and automation [13], [14], [15]. This shift is accompanied by increased skill-based competition, where individuals with AI-related expertise are in high demand, fostering a competitive job market that rewards advanced skills and knowledge [1], [3], [4], [6].

Evidence from the US, the UK, OECD countries, and emerging markets shows a notable rise in the demand for AI skills, accompanied by significant wage premiums for those with AI expertise, particularly in high-skilled occupations and large firms [1], [3], [7], [8], [10]. Importantly, AI technologies often complement, rather than replace, employees [10]. This synergy enhances overall productivity and efficiency, particularly in the digital transformation process across various industries [4], [10].

In the US, the introduction of generative AI has been associated with increased employee productivity by reducing the average time spent on tasks such as writing and customer service, while also enhancing the quality of output [16]. These improvements are more pronounced for less-experienced and lower-skilled employees than for their more experienced and highly skilled counterparts [16], [17]. These patterns suggest that generative AI helps newer employees progress more rapidly along the learning curve and reduces productivity inequality among employees [16], [17]. Furthermore, the use of generative AI has been linked to improved employee retention, particularly among newer staff members [16].

In China, AI reduces production costs, lowers prices, and increases employees' incomes, which stimulates demand and consequently drives economic growth [4]. In the same region, AI drives innovation, leading to new products, models, and industrial sectors [4]. Moreover, AI contributes to broader economic shifts through virtual agglomeration, enabling businesses and employees to connect and collaborate across different geographical locations [4]. This fosters a global network of employment opportunities, enhancing global connectivity and supporting diverse job roles across various industries.

Additionally, AI helps industries integrate advanced technology into business operations, making them more resilient during economic downturns by optimizing processes, reducing operational costs, and opening new revenue streams [2]. In China, this resilience has contributed to stabilizing employment levels during challenging financial times [4]. Moreover, during the COVID-19 pandemic in the UK, AI helped SMEs mitigate workforce-related business risks [2]. By enhancing efficiency and rapidly pivoting business operations, AI allowed SMEs to navigate the complex financial challenges presented by the COVID-19 pandemic [2].

AI enhances the quality of employment, particularly for historically disadvantaged groups like female employees [4]. In China, AI-driven productivity improvements have led to more flexible, safer, and more fulfilling work environments [4]. With the rise of automation, robotics, AI, and advanced machinery, the importance of physical strength has diminished. Instead, there is a growing emphasis on cognitive abilities, problem-solving, creativity, and emotional intelligence—skills that are not bound by gender. These advancements have helped bridge the gender gap in traditionally male-dominated industries by reducing reliance on physical strength and increasing the importance of cognitive and emotional skills.

In the UK, Greece, and Cyprus, training on the use of advanced technology has been found to reduce gender and ethnic digital divides and boost empowerment in the labor market by enhancing digital competencies and addressing intersectional barriers faced by immigrant women, including socioeconomic and cultural conditions that exacerbate digital disparities [18], [19]. Similarly, in OECD countries, AI in high-tech operations has been associated with better job quality because it complemented highly sophisticated job tasks [8].

AI and drawbacks in the labor market

Review studies have shown that AI has widened the gap between high- and low-skilled employees, particularly by increasing the risks of job displacement for low-skilled individuals and those in routine-based occupations due to automation [13], [14], [15]. US firms that invest more in AI tend to transition towards more highly educated workforces [12]. Specifically, these firms see an increase in the proportion of employees with undergraduate and postgraduate degrees, particularly in STEM fields, while the share of employees without a university degree decreases in AI-investing firms [12]. Medium-skilled roles are in decline, suggesting that AI may substitute certain non-technical positions [12].

In both advanced economies and emerging markets, employees in low-complementarity occupations are more vulnerable to job displacement due to high exposure to AI [10], [14], [15]. This shift has the potential to put downward pressure on wages in occupations exposed to AI that do not require advanced skills [8]. Even when AI enhances the productivity of low-skilled employees, it may not reduce wage inequality [11]. Instead, those who own AI technologies or invest in them would see higher returns, while other employees could face stagnant wages or job displacement. This contributes to exacerbating overall inequality [11].

Employees often perceive AI as a threat to their jobs, particularly those in roles involving repetitive tasks, data management, and routine physical operations [20]. When AI systems fail to meet employees' expectations, trust in these technologies may diminish. This distrust could intensify if firms exaggerate AI’s capabilities, leading employees to perceive AI as a threat to their job security, particularly in roles susceptible to automation. Furthermore, a symbiotic relationship between employees and AI systems might be difficult to achieve if employees do not develop the technical, human, and conceptual skills needed to coexist with AI [20]. The use of AI in performance evaluations, hiring, and promotions can seem impersonal and biased, especially when employees are not provided with clear explanations of how AI systems make decisions. This sense of alienation is further compounded when employees feel that their roles are increasingly dictated by algorithms rather than human judgment, leading to increased stress and anxiety among employees [8].

Additionally, there are concerns about AI's ethics and transparency, especially in decision-making processes, which can erode trust in management [8], [20]. As AI systems increasingly rely on vast amounts of data to function, concerns about how this data is collected, stored, and used have come to the forefront. Employees are often uneasy about the extent to which their personal data is being monitored and analyzed, leading to fears about privacy and potential misuse of information. These concerns are not only related to personal privacy but also to job security, as employees worry that data-driven insights could be used to justify downsizing or restructuring decisions [8].

Theoretical background

The studies reviewed in this paper have employed a combination of theoretical frameworks to evaluate the relationships between AI and labor market outcomes at the macro level. Several studies have utilized Schumpeter’s Innovation Theory to explain how AI and automation disrupt existing jobs while creating new ones, leading to shifts in employment patterns [1], [5]. Additionally, the Technology-Task Substitution and Complementarity Theory has been used by some researchers to examine how AI either substitutes human labor—leading to job displacement—or complements it, enhancing productivity and creating new jobs [5], [6], [8].

In the reviewed literature, Innovation, Growth, and Diffusion Theories have been applied to explain how AI drives firm and industry growth through technological innovation and institutional factors, contributing to economic expansion and labor market outcomes [3], [8]. Many studies have utilized the Skill-Biased Technological Change framework, which examines how AI increases the demand for high-skilled labor while reducing the demand for low-skilled labor, leading to wage inequality [1], [5], [8], [14]. A key component of this framework is the Labor Market Polarization Theory, which addresses the division of the labor market into high-skill, high-wage jobs and low-skill, low-wage jobs due to technological advancements like AI, often at the expense of middle-skill jobs [5], [6], [8], [10], [11], [14].

At the micro level, Human Capital Theory has been applied in the literature to examine the importance of education, skills, and adaptability in enabling employees to benefit from AI rather than being displaced by it [1], [3], [5], [6], [8]. New micro-level frameworks have also been proposed to capture how AI shapes individual realities. For instance, the AI capital framework explores how knowledge, skills, and capabilities related to AI technologies can enhance an individual's value in the labor market [1], [18]. Additional frameworks include the Business Apps Training framework, which suggests that training individuals in business applications on advanced technology can increase digital competencies, empowerment, and reduce digital divides [18], [19]. Moreover, the Business AI Apps framework examines how firms can use advanced technology to overcome financial challenges that affect the workforce [2].

A synthesis of macro and micro-level models is crucial in this domain because outcomes observed at the macro level, such as changes in labor market polarization, are directly influenced by micro-level factors like individual education and skill acquisition. Conversely, individual decisions and behaviors aggregate to create the broader trends studied in macro-level models. Understanding this interplay is essential for accurate predictions and effective policymaking. For example, macro-level insights might suggest the need for large-scale educational reforms, while micro-level analyses can inform targeted training programs that address specific skill gaps, ensuring that policies are both comprehensive and effective [1], [18], [19].

Limitations and gaps

While some regions, such as the US and parts of Europe, are well-studied, there is a lack of comprehensive research covering a wider range of countries, particularly in the Global South [14], [15]. Research on AI's impact employs various proxies for AI exposure, such as robotization, digital evolution indices, and vacancy-based measures [3], [7]. While these proxies are useful, they often lead to diverse and sometimes contradictory conclusions on AI’s impact on employment and growth [8], [9], [11].

Moreover, the effectiveness of AI in job creation and productivity is complex and multifaceted, with significant variations observed across different regions indicating that the assigned patterns are not homogeneous [8], [10]. This regional variation suggests that AI's benefits and drawbacks are not uniformly distributed and are influenced by local economic conditions, regulatory environments, political realities, and industry characteristics [13], [14]. The literature also lacks a clear consensus on which skills or sectors will be most affected by AI [3], [4], [7], [8]. 

Additionally, there is a scarcity of robust empirical evidence on the long-term effects of AI on employment, wages, and economic growth [8], [9], [11], [14]. Most studies focus on short-term impacts, leaving a limited understanding of how AI might reshape labor markets over time, particularly across different sectors and regions [14], [15] .

Furthermore, much of the literature examining the economic impacts of AI often overlooks the ethical, social, and psychological implications for employees, such as increased monitoring, job stress, and the potential for decreased job satisfaction due to AI-driven changes in the work environment [8], [14], [15]. 

In addition, the literature on AI's impact tends to focus heavily on macro-level evaluations, often at the expense of exploring micro-level behaviors and outcomes [1]. For instance, how AI capital directly influences employment outcomes at the individual level remains underexplored. This gap largely stems from the lack of data capturing individuals' AI capital levels. Without this micro-level data, it is challenging to assess how AI might be reshaping job roles, altering skill requirements, and influencing individual career trajectories [1].

Summary and policy advice

This study shows that AI has boosted employment, particularly in high-skilled roles and large firms [1], [3], [4], [6]. Moreover, AI has also led to wage premiums and improved business operations in high-skilled occupations [1], [3], [7], [8]. Furthermore, AI has been shown to enhance productivity for less-experienced and lower-skilled employees by reducing the average time spent on tasks, improving the quality of output, stabilizing employment during economic downturns, and enhancing job quality for historically disadvantaged groups [2], [4], [8], [16], [18], [19].

However, the widespread adoption of AI has increased the risk of job loss for low-skilled employees, with fears of widening income inequality [11], [14], [15]. Additionally, AI adoption has disrupted workplace trust and raised concerns over ethics [8], [20] .

To maximize the benefits of AI while mitigating its adverse effects, strategic interventions are essential [1], [8]. Governments should play a proactive role by significantly increasing investments in AI research and development (R&D), ensuring that these innovations align with societal needs and values [2], [4]. By funding R&D, governments can drive breakthroughs in AI that lead to new industries and job creation, fostering long-term economic growth [2], [4].

Moreover, it is imperative to implement robust policies that focus on workforce reskilling and upskilling [1], [18], [19]. These training programs should include not only technical skills, such as data analysis and programming, but also soft skills like critical thinking, creativity, and adaptability, which are increasingly important in a world where AI handles routine tasks [1]. By doing so, these policies will help displaced individuals transition into new employment opportunities, thereby reducing concerns about AI's impact on job security [1], [5]. In addition, governments should make strategic investments in education and training programs aimed at improving digital literacy and AI capital among the general population [1], [18]. This will not only enhance their employability but also contribute to overall productivity and economic resilience [1], [18].

Policymakers should also develop region- and sector-specific strategies to address the unique challenges posed by AI [5], [14], [15]. For instance, regions heavily reliant on manufacturing or other industries highly susceptible to automation may require targeted investments in digital infrastructure and specialized training programs to mitigate the risk of significant job displacement [5]. Moreover, enhancing social security systems, including unemployment insurance and public welfare jobs, is crucial for managing the structural unemployment caused by AI-driven automation [4], [14]. By providing a safety net for displaced employees, these systems can help maintain social stability and prevent the exacerbation of income inequality [4], [14].

AI adoption strategies should be inclusive, ensuring that the benefits of AI are broadly distributed across all segments of the workforce, particularly among vulnerable groups such as low-income employees, minorities, and those with less access to education and training opportunities [18]. This can be achieved by encouraging AI adoption in ways that complement human labor rather than replace it [10].

Safeguarding employee rights in the context of AI adoption is crucial, including ensuring fair wages, preventing discrimination, and ensuring that individuals' personal information is protected in AI-driven processes [8]. By establishing these ethical standards, policymakers can ensure that AI technologies are implemented in a manner that benefits society as a whole while minimizing potential risks [8], [20]. Transparency in AI decision-making is essential for building trust among users and ensuring that AI systems are accountable [8], [20].

Acknowledgments

The author thanks three anonymous referees and the IZA World of Labor editors for many helpful suggestions on earlier drafts. The author’s previous work contains a larger number of background references for the material presented here and has been used extensively in all major parts of this article [1], [2], [18], and [19].

Competing interests

The IZA World of Labor project is committed to the IZA Guiding Principles of Research Integrity. The author declares to have observed these principles.

© Nick Drydakis