AI and Labour
The evolution of automation and its impact on labour is a narrative of technological advancements intertwined with economic transformations and societal shifts. Beginning with the Industrial Revolution, automation expanded as machinery replaced manual labour in factories, heralding unprecedented productivity gains but also sparking concerns about job displacement. Mass production techniques in the early 20th century further accelerated this trend, epitomized by the advent of assembly lines, marking a significant juncture where efficiency soared but traditional craft-based occupations faced irrelevance.
The latter half of the 20th century witnessed the digital revolution, propelling automation beyond manufacturing into diverse sectors with the emergence of computer-controlled systems. This era saw the onset of globalization, coupling automation with outsourcing, altering the landscape of labour markets worldwide. The 21st century saw the rise of new automation and digital technologies, including AI, revolutionizing the way we work and shaping our economy.
AI, along with other automation tools, can perform tasks previously done by humans, often quicker, cheaper, and more efficiently. Many people rely on income from work to sustain their livelihoods, so it is essential to understand how AI, like other forms of automation, impacts employment. One crucial concern is how AI will affect the availability of good jobs that offer fair wages, safe working conditions, legal protections, benefits, and opportunities for career growth. In societies where access to such jobs is vital for stability and well-being, this is especially important.
While some believe that with the right economic policies, there will be enough jobs to match the slow growth of the labour force in advanced economies, others are sceptical. They doubt that AI and ongoing automation will generate sufficient quality jobs or distribute productivity and growth benefits fairly, possibly increasing income and wealth inequality.
However, there is optimism that strategic interventions can mitigate AI's negative impacts on employment. Policies such as social insurance, education and training programs, and tax measures, along with institutional mechanisms like collective bargaining, can play a role in shaping the effects of AI on labour.
Contemporary AI relies on powerful computing, vast datasets, and sophisticated algorithms to automate tasks previously requiring human cognitive abilities. Despite concerns about its impact on employment, there is potential for AI to enhance productivity and improve our lives if managed wisely and with consideration for its social and economic implications.
Machine learning (ML) advancements in recent years have led to breakthroughs in Artificial Intelligence, enabling AI systems to perform tasks that rival human abilities in various areas. However, these AI applications, known as narrow AI, are still limited in scope and can only handle specific tasks without the ability to transfer knowledge from one problem to another. While narrow AI can take over repetitive, data-heavy, and optimization-based tasks, it falls short in tasks requiring high-level cognitive abilities such as reasoning, judgment, and social interactions.
Narrow AI is also impacting human tasks by enhancing robots and production systems with intelligence and powering digital platforms that facilitate transactions between users. This creates a cycle where AI collects, analyses, and predicts data, fuelling the growth of platforms like Amazon, Netflix, Uber, and Upwork, and so on which provide various goods and services.
Predictions about the future effects of AI are filled with uncertainties, including how quickly and broadly AI will be adopted and what scientific breakthroughs will occur. While scientific progress determines if a task can be automated, decisions about whether to automate tasks depend on various factors, such as business strategies and market conditions.
For example, businesses often invest in automation technologies when taxes on labour are high compared to taxes on machinery and software. However, the deployment of these technologies is gradual due to the time needed to develop the necessary organizational capabilities.
AI is expected to continue, intensify, and speed up the adverse effects of automation on labour markets in advanced economies. This includes trends like job polarization, slow wage growth for middle- and low-skilled workers, higher wages for highly educated workers, wage growth not keeping pace with productivity, a decrease in labour's share of value-added, and growing income inequality.
Automation isn't the only factor behind these trends. Globalization, outsourcing, declining unionization, and the increasing bargaining power of businesses also play significant roles. Automation, in turn, has enabled or reinforced these factors. For instance, robots and digitalization have accelerated globalization by making it easier to outsource routine jobs to low-wage countries, reducing employment and limiting wage growth in manufacturing and tradable services in advanced economies.
While advanced countries have all faced similar technological and globalization pressures affecting labour, there are notable differences in outcomes among them. This disparity is partly due to variations in policies, institutions, and societal norms regarding fairness. Countries like Germany, Sweden, Canada, and Denmark, known for their strong economies, have managed these forces more effectively than the United States, resulting in better outcomes for their workers.
However, automation disrupts the labour market, creating winners and losers. Short-term trade-offs exist for businesses, workers, citizens, and political leaders, with economic, social, and political consequences. Moreover, history is replete with evidence that the social and political costs of labour market disruptions triggered by technological change can be significant.
The job displacement effects of automation can impact specific areas while productivity and job creation benefits may occur elsewhere. This can complicate politics and policies, as costs and benefits are often unevenly distributed.
For the past few decades, automation has sped up and intensified, but the expected increases in productivity and job creation have been slower and smaller. This has led to growing social and economic disruptions, with the benefits not being as widespread as anticipated.
Much of the automation in recent decades has focused on routine tasks, both manual and cognitive. These routine tasks, such as those in manufacturing and office work, have seen a decrease in employment due to automation.
AI is often categorised as "routine-biased technological change" (RBTC) because it replaces humans in routine tasks while increasing the demand for non-routine tasks. AI enhances automation by adding intelligence, particularly in routine physical and cognitive tasks. This includes tasks like assembly-line production and customer support, which can be automated due to their routine and data-intensive nature.
Tasks that involve high-level thinking and require social interaction, creativity, and complex strategies, like those performed by professionals in business, healthcare, education, and the arts, are not directly targeted by current AI technologies.
If we think of AI as a supercharged version of past technological advancements, its impact on jobs will likely resemble the effects seen over the last few decades with other types of automation. One major impact is the "polarization" of employment and wages. Many jobs lost to automation in the past were in manufacturing, which used to provide good opportunities for middle-skill workers. This polarization means fewer jobs in the middle-skill range, with more in both low-skill and high-skill categories, especially the latter. Despite job loss in the middle, there's been an increase in higher-skilled positions overall.
This polarization has led to wider wage gaps among workers. Those in jobs displaced by automation have seen little to no wage growth, while those benefiting from productivity gains or new job creation have experienced wage increases. This has contributed to growing income inequality across developed countries, driven by the growing pay disparity between highly educated workers, whose skills are complemented by automation, and those with lower levels of education or training, whose skills have been replaced by automation.
While acknowledging the potential for AI to enhance human productivity in certain tasks and even create new roles requiring uniquely human skills, it's important to recognize the distinct abilities that humans possess, which AI cannot replicate. These include social and interpersonal skills crucial in professions like teaching, caregiving, healthcare, physical therapy, and hairstyling. Additionally, humans excel in tasks requiring physical dexterity in unpredictable environments, such as construction and plumbing, as well as in nonroutine problem-solving and general intelligence needed in management and artistic endeavours.
Looking ahead, the future of work is likely to see a symbiotic relationship between human skills and AI capabilities across various occupations. For instance, doctors and teachers may increasingly rely on AI for data analysis, diagnostics, and predictive insights, while still leveraging their interpersonal abilities. This interdependence between human and AI skills is anticipated to result in the emergence of partnership-based occupations, necessitating advanced education and technical training for human workers. However, this shift may also increase wage disparities, as workers whose skills are augmented by AI could command higher incomes compared to those whose roles are displaced.
A critical yet unresolved question revolves around the distribution of rewards from labour between humans and their AI counterparts, as well as between workers and the individuals or entities that own and develop these intelligent tools. As AI becomes more integrated into various aspects of work, addressing these issues becomes necessary to ensure equitable outcomes for all stakeholders involved.
Up to this point, we discussed how AI is reshaping labour demand by automating various tasks and occupations. Now, let's look at how AI is influencing labour through the emergence of digital platforms, which are creating new tasks and forms of organizing work.
The use of digital platforms, particularly during the COVID-19 pandemic, has seen a significant surge and is expected to continue expanding rapidly. To anticipate AI's future impact on labour, it is crucial to examine this phenomenon through the lens of digital platforms. AI is enabling the development of mainly three types of digital platforms:
1. Platforms for selling goods: Companies like Amazon and Netflix are redefining which tasks are performed by humans and where they are performed. With the acceleration of e-commerce and digital transactions due to COVID-related shifts in business practices, tasks are transitioning from traditional retail settings to warehouse operations and various stages of delivery and transportation.
2. Platforms for labour services: Platforms such as Upwork, Lyft, and TaskRabbit are leveraging algorithms and real-time data to connect workers with tasks across a wide spectrum of industries. These platforms cover tasks ranging from nonroutine cognitive work to routine personal services, often on a temporary or project basis. This type of work, commonly known as "gig work," includes roles like digital assembly-line workers, who provide human intelligence for AI software, and is a significant component of the on-demand gig economy.
3. Platforms for renting out assets: Companies like Airbnb and BlaBlaCar offer opportunities for individuals to earn income by renting out assets like accommodations or vehicles. While these platforms create new opportunities for labour and income, they also reshape the nature of work and the skills required for tasks.
The growth of platform-mediated work is rapidly increasing as a share of nonstandard employment arrangements, which already account for a significant portion of the working-age populations in advanced economies. While the platform-mediated gig segment remains relatively small as a percentage of total employment, it is expanding swiftly, reflecting broader trends toward nontraditional work arrangements.
On the other hand, Gig workers find themselves in a precarious position, lacking the legal and social protections afforded by traditional employment contracts. This situation leaves them vulnerable to various challenges, including unstable incomes, limited access to social insurance, few opportunities for training and career advancement, exposure to health and safety risks, and minimal collective bargaining rights.
Looking ahead, the future of labour will be influenced by two major factors: the demand for goods and services and the capabilities of intelligent tools and systems driven by AI. These forces are expected to reshape employment trends in advanced industrial economies over the next decade. Occupations involving routine tasks, such as office support, production, and warehousing, are likely to decline in favour of roles in sectors like healthcare, education, technology, and the arts, which require nonroutine tasks.
The trend of upskilling in the labour market is expected to persist, with job growth concentrated in high-wage occupations while lower- and middle-wage jobs may see declines. This polarization of the labour market is likely to increase wage inequality. Additionally, the transition for workers displaced by AI and automation will come with significant costs. These individuals will need to acquire new skills for emerging roles, leading to questions about who should bear the burden of these transition costs.
The advancements in automation have also led to substantial increases in productivity, but these gains have not been evenly distributed among workers. Instead, a disproportionate share has flowed towards capital, leaving workers with stagnant wages and a shrinking share of the value added. This trend is evident in the widening gap between productivity growth and wage growth, as well as the declining proportion of labour's contribution to overall value added. Particularly hard-hit by these shifts are workers in the lower half of the earnings distribution, those without a college degree, and regions where manufacturing historically played a significant economic role.
As AI continues to drive automation in manufacturing, similar disparities in the distribution of benefits and costs are expected to persist. While the gradual depletion of manufacturing jobs is anticipated to continue, it is projected to be less severe than in previous decades. The emergence of AI-powered automation will introduce increasingly programmable and interconnected machines, optimizing production systems by automating existing tasks and creating new complementary tasks that require new skills from workers.
The pace of automation in manufacturing will be influenced not only by advancements in AI capabilities but also by improvements in the dexterity of robots and production systems. While AI is not expected to significantly increase the risk of job displacement for shop-floor manufacturing workers, routine cognitive tasks in back-office settings are likely to be affected. However, the demand for services is expected to remain high in advanced industrial societies, driven by rising incomes and changing demographics. The service sector already accounts for the majority of employment and employment growth, encompassing a wide range of occupations from highly paid professionals to middle-wage educators and low-wage retail and hospitality workers.
Two major service industries, retail and health, are particularly impacted by AI. In the retail sector, AI technologies are revolutionizing customer experiences through personalized recommendations, inventory management, and cashier less checkout systems. In healthcare, AI is transforming diagnostics, treatment planning, and patient care, improving efficiency and outcomes. While AI brings opportunities for innovation and efficiency in both sectors, it also raises concerns about job displacement and the need for upskilling and retraining programs to ensure workers can adapt to the changing landscape.
Moreover, AI applications are poised to replace humans in cognitive tasks reliant on data across administrative and office support roles and patient relationship management. Concurrently, these advancements are expected to heighten the need for human involvement in healthcare professions such as nursing, medicine, physical therapy, and dentistry, where tasks necessitate high-level cognitive abilities and skilled social interaction. The integration of AI into administrative and data collection tasks, bolstered by telemedicine platforms, stands to revolutionize nursing practices, allowing nurses to dedicate more time to direct patient care by leveraging AI-generated insights for real-time health guidance, diagnosis, and treatment.
The future landscape of healthcare will likely entail collaborative efforts between human professionals equipped with essential social skills and intelligent tools armed with robust data capabilities. This collaborative model is already manifesting in the healthcare sector through the deployment of AI-enabled robots, particularly in countries like Japan, which are grappling with ageing populations and a shortage of healthcare workers. These robots are being employed in nursing homes and hospitals to address labour shortages and complement human staff, thereby optimizing the delivery of care services.
So far we have seen how AI and automation impact the demand for human labour across various tasks, occupations, and industries. The evolving dynamics underscore the need for a nuanced understanding of how AI integration influences the composition of the labour market, necessitating strategic responses to ensure a balanced and sustainable approach to workforce development and deployment in the era of AI technology.
As economies around the world recover from the impacts of COVID-19, concerns are emerging in the United States and several European countries regarding the future availability of workers possessing the necessary skills and education to meet the demands of expanding sectors such as healthcare and software engineering. These anticipated shortages are expected to drive further innovation, investment, and implementation of AI-enabled automation technologies as substitutes for human labour.
The advancement of AI and the intelligent tools it enables will lead to the automation of numerous routine tasks, alterations in existing tasks, and the creation of entirely new tasks for human workers. This transformative process will also involve novel forms of collaboration between humans and machines, as well as new organizational structures within workplaces.
However, It cannot be assumed that the transformation of work by AI will inherently result in broad economic gains that are equitably shared across society. Rather, achieving such outcomes requires deliberate societal choices and policy interventions. To foster economic growth alongside social and economic equity, advanced economies must develop policies aimed at distributing both the costs of disruption and the productivity benefits of AI in a manner consistent with prevailing norms of fairness and justice.
Ensuring the availability and accessibility of quality jobs is crucial policy objectives, though achieving them presents significant challenges. To enhance the likelihood of success and elevate all job opportunities to a high standard, three broad categories of policy interventions are necessary. Firstly, lifelong education and training policies are essential to equip workers with the necessary skills for good job opportunities, accompanied by active labour market policies to assist their transition into these roles. Secondly, extending social benefits and legal protections to encompass workers across all industries, including platform-based businesses, is important. Lastly, a combination of income-support policies, such as minimum wages, work tax credits, and basic income supplements, is needed to elevate the earnings of workers in low-wage positions, particularly those in routine service sectors like leisure, hospitality, healthcare, and childcare, where women and low-educated workers are prevalent.
It's crucial to emphasize that the impact of AI on employment is not solely determined by technology but is influenced by the motivations of both AI researchers and investors in its deployment. The prevailing narrative in the business and research sectors, especially pronounced in the United States, prioritizes AI's ability to surpass human capabilities rather than focusing on job creation. This narrative is bolstered by tax policies that favour capital over labour, encouraging businesses to adopt automation technologies that reduce employment and labour costs without commensurate gains in productivity. Government support for research in labour-saving technologies further reinforces this narrative, but well-crafted policies have the potential to shift it.
Ultimately, the realization and equitable distribution of the economic benefits stemming from intelligent machines rely not on their technological features but on the formulation of intelligent policies essential for an inclusive AI era to flourish.
The evolution of automation and its impact on labor spans centuries and encompasses profound technological advancements, economic transformations, and societal shifts. Beginning with the Industrial Revolution, automation burgeoned as machinery replaced manual labor in factories, leading to unprecedented productivity gains but also raising concerns about job displacement. The introduction of mass production techniques, epitomized by assembly lines in the early 20th century, further accelerated this trend, marking a significant shift where efficiency soared while traditional craft-based occupations faced irrelevance.
The latter half of the 20th century witnessed the digital revolution, extending automation beyond manufacturing into various sectors with the rise of computer-controlled systems. This era also saw the onset of globalization, intertwining automation with outsourcing and reshaping labor markets worldwide. As we entered the 21st century, the proliferation of AI and robotics ushered in a new era of automation, promising unparalleled efficiency while sparking debates about job displacement and the need for workforce adaptation.
AI, alongside other automation tools, has the capability to perform tasks previously done by humans, often quicker, cheaper, and more efficiently. This raises essential questions about the availability of quality jobs that offer fair wages, safe working conditions, and opportunities for career growth. Concerns persist regarding whether AI will generate enough quality jobs and distribute productivity and growth benefits fairly, potentially exacerbating income and wealth inequality.
However, there is optimism that strategic interventions, such as social insurance, education and training programs, and tax measures, along with institutional mechanisms like collective bargaining, can mitigate AI's negative impacts on employment. Despite concerns, there is recognition of AI's potential to enhance productivity and improve lives if managed wisely, emphasizing the importance of navigating its implications and formulating intelligent policies for an inclusive AI era.