The Evolution of Job Displacement in the Age of AI and Automation: A Bibliometric Review (1984–2024)

1.1 Study Objectives

• To analyse the scientific production related to AI, automation, and job displacement over 40 years (1984–2024).

• Using bibliometric tools to identify trends and patterns in annual scientific output on AI and job displacement.

• To evaluate the impact of published works through citation analysis and average citations per year.

• To determine the most influential journals and sources publishing research on AI, automation, and workforce displacement.

• To identify key authors and institutions contributing to research on job displacement and automation.

• To assess international collaboration trends in AI and job displacement research.

• To analyse core sources of AI and job displacement research through Bradford’s law.

2.1 Theoretical Framework on Job Displacement Due to Technological Advancements

Technological advancements have long been viewed as a threat to human labour, as seen during the first Industrial Revolution when the introduction of power looms and mechanical knitting frames led to the Luddite movement, which opposed this shift (Mokyr, Vickers, & Ziebarth, 2015). During the nineteenth century, prominent economists Karl Marx and David Ricardo supported the idea that technology could render workers obsolete (Keynes, 2010). According to research, technological advancements are disrupting labour markets, leading to a growing polarization of employment opportunities into low-skilled and high-skilled positions (Cirillo, Evangelista, Guarascio, & Sostero, 2021). The rapid advancements in robotics and AI are anticipated to significantly alter the labour market (Katz & Murphy, 1992). The rapid growth of new technologies raises concerns about the potential displacement of high-skill, high-wage jobs, potentially putting previously secure positions at risk (Lloyd & Payne, 2019). By 2025, these transformations will create 97 million new jobs and eliminate 85 million existing ones globally (Susskind, 2020). Studies predict significant job losses in automation-prone roles, influencing on-going discussions on the impact of technological advancements on the labour market (Arntz, Gregory, & Zierahn, 2017; Frey & Osborne, 2017). Recent studies have challenged this perspective, highlighting the positive effects of technology on employment (Aghion, Antonin, Bunel, & Jaravel, 2022).

The theoretical framework on job displacement due to AI and automation draws on multiple essential perspectives from economics, sociology, and labour theory. Central to this framework is the theory of skills-biased technological change; technological advancements are believed to enhance the necessary for skilled labour, leading to wage inequality, the polarization of the workforce (Bolli & Pusterla, 2023), and increased wages (Card & DiNardo, 2002). The theory suggests that technology boosts the productivity of highly skilled jobs, replacing low to medium-skilled employment, leading to an increase in income inequality (Cirillo et al., 2021). Complementing this is routine-biased technological change, which explains how AI and automation replace human labour in routine tasks, contributing to the decline of mid-skilled jobs (Autor, Levy, & Murnane, 2003; Goos & Manning, 2007), and highlights the disappearance of the “middle” in manual and cognitive jobs, which machines can replace due to predictable and routine tasks (Goos & Manning, 2007). This displacement of routine work is further contextualized by Schumpeter’s creative destruction, wherein technological innovations disrupt traditional industries, eliminating specific jobs while creating new opportunities (Susskind, 2020). Marxist labour theory also offers critical insights, emphasizing how technological advancements can generate surplus labour and exacerbate socioeconomic inequality, mainly as low-skill jobs disappear and high-skill jobs become more secure (Keynes, 2010).

2.2 Examination of the Relationship Between AI/Automation and Labour Market Dynamics

The interplay involving AI, automation, and labour market intricacies is continuously changing. This relationship substantially affects productivity levels, job displacement, income disparity, workforce competencies, and the overall economic framework. Technological advancements have varying societal impacts, with some arguing that they lead to job losses while others argue that they do not. This perspective is rooted in labour sociology and economics, but historical evidence since the Industrial Revolution suggests otherwise (Zhang, 2023). The rapid progress of AI technologies drives industrialization and commercialization processes while the industry continues exploring its application across various domains. AI is revolutionizing intelligent manufacturing, fostering a high-quality employment model based on effective human-machine collaboration, thereby generating and fulfilling job opportunities (Shen & Zhang, 2024).

Robots are often seen as rivals to humans, but this perspective is materialistic. The relationship between humans and machines should not be zero-sum. Focusing on collaboration can alleviate production limitations, enhance productivity, and create more employment opportunities and innovative roles rather than viewing robots as rivals (Duan, Deng, & Wibowo, 2023). Implementing materialized AI technology can improve production efficiency, aligning with the existing factor endowment structure. It can optimize production between upstream and downstream enterprises, facilitating production scaling and reproduction. This synergy fosters growth in labour demand across various skills, leading to a creative impact and concurrent increase in labour demand (Liu, Gu, & Lei, 2022). AI, a vital component of the fourth industrial revolution, significantly impacts human social status and workforce composition (Chen, 2023). AI and machinery are revolutionizing labour productivity by automating repetitive tasks, enhancing employee skills, and increasing the value of work. This shift may lead to the disappearance of low-skilled positions and new job opportunities (Polak, 2021).

AI has both a positive and negative impact on employment, as it streamlines complex tasks and represents a human-like intelligence. Like early industrial technologies, automation technologies like AI present opportunities and concerns about machine replacement. The rise in technology leads to increased capital composition and a surplus population, highlighting the complex relationship between technology and employment (Shen & Zhang, 2024). Artificial intelligence has significantly enhanced productivity in healthcare, transportation, and production environment management, leading to increased leisure time (Shen & Zhang, 2024). Despite the widespread use of AI, several industrialized nations have experienced declining labour income and slower productivity growth (Autor, 2019). Workers with low skills and those who are incapacitated are at a high risk of being replaced by automation (Ramos, Garza-Rodríguez, & Gibaja-Romero, 2022). AI and robotics significantly impact the manufacturing job market, leading to a significant unemployment issue in industry-related positions due to their disruptive nature (Bian, 2024).

2.3 Empirical Studies on AI and Job Displacement

Job displacement due to automation often results in individuals transitioning from stable, high-quality employment to less secure, lower-quality roles in the secondary labour market. This shift can lead to diminished job security, limited social mobility, and a decline in social standing. Over the past years, numerous studies have examined the impact of automation on employment, revealing significant concerns about job displacement.

A study by Frey and Osborne (2017) suggests that AI could replace 47% of the 702 US job categories within the next two decades, particularly as digitization accelerates post-pandemic. Similarly, research across various countries indicates that automation is poised to significantly impact employment. For instance, 60% of jobs in Brazil are expected to be affected by automation in the coming decades, with eight out of the ten most common occupations being highly susceptible (Lima et al., 2021). In the UK, 35% of jobs are projected to be substantially impacted (Frey & Osborne, 2014), while in Canada, 42% of jobs are at risk (Lamb, 2016).

The influence of technology on professions is also evident. A survey of technology professionals found that 48% believe robots can effectively perform routine tasks (Ngo et al., 2014). The rise of AI is expected to exacerbate high-tech unemployment by transforming traditional industries into routine tasks that technology will eventually replace (Susskind & Susskind, 2016). Research indicates that 55% of jobs at high risk of AI substitution face no gender disparity (David, 2017). Additionally, the introduction of robots has significantly reduced the responsibilities of low-skilled labour compared to medium-skilled and high-skilled labour (Graetz & Michaels, 2018). The increased use of robots in US labour markets has led to a 0.2% decrease in the employment-to-population ratio and a 0.42% decrease in wages (Acemoglu & Restrepo, 2020).

A study by Arntz, Gregory, and Zierahn (2016) found that, on average, 9% of jobs in 21 Organization for Economic Cooperation and Development (OECD) member nations face significant automation risk. This risk varies across countries, ranging from 12% in Germany to 6% in Estonia (Arntz et al., 2016). A broader analysis of 32 OECD nations revealed that 14% of jobs are highly susceptible to automation, with a 70% probability, while 32% face a 50–70% automation risk (Nedelkoska & Quintini, 2018).

Despite these concerns, several studies highlight the potential benefits of AI implementation. A study on a Spanish manufacturing company found that adopting robots led to a 10% net job increase, while firms that avoided investing in robotic technology experienced job losses (Koch, Manuylov, & Smolka, 2021). Graetz and Michaels (2018) also found that the use of robots increased annual labour productivity by 0.36 percentage points, positively impacted total factor productivity, and decreased output prices, although it did not significantly affect overall employment levels except for low-skilled workers. Nobel laureate Boden (1987) argues that the advancement of AI creates new jobs that surpass those lost, making long-term unemployment concerns unwarranted. Researchers suggest that while AI may initially impact the job market, its enhanced production efficiency will ultimately lead to increased production and employment opportunities (Guliyev, 2023).

2.4 Workforce Adaptation Strategies, Including Reskilling and Upskilling Initiatives

AI is not meant to substitute human labour wholly but to complement and improve human abilities, ensuring a smooth transition and fostering growth in the workforce (Idrisi, Geteye, & Shanmugasundaram, 2024). AI can automate roles previously performed by humans, potentially leading to job displacement in specific sectors (Idrisi et al., 2024). Automation in certain positions presents new opportunities and increased requirements for skills in developing sectors. The integration of AI requires the acquisition of the necessary skill sets (Baral, Rath, Goel, & Singh, 2022), such as education and training by utilizing innovative educational approaches, emphasizing experiential learning, and addressing a range of competencies, encompassing essential technical, leadership, and digital technology implementation skills (Babashahi et al., 2024). Workers must develop digital skills, higher education, technical expertise, and strong literacy and numeracy skills to adapt to technological advancements and reduce the risk of automation replacement (Filippi, Bannò, & Trento, 2023).

The industry domain (Jurczuk & Florea, 2022) emphasizes the importance of digital skills in future manufacturing, including process design, automation, creativeness, critical imagining, digital material development, data assessment, and cybersecurity. In the legal domain, law students need to understand AI to integrate legal expertise with technological insights, advocating for AI-focused modules in academic programs and on-going professional development opportunities (Beebeejaun & Gunputh, 2023). Chirgwin (2021) emphasizes the significance of integrating technical and emotional competencies in mine controllers, stressing the necessity for a blend of hard and soft skills in AI adoption. Meanwhile, Kumar, Naveen, Kumar Illa, Pachar, and Patil (2023) emphasize the importance of software engineers continuously enhancing their skills in AI tools like language processing, machine vision, machine learning, big data, and IoT technologies to stay applicable in the progressively developing AI landscape. Further, Faraj (2022) highlighted the relevance of developing technical competencies in digital, technological, scientific, mathematical, engineering, and programming fields and soft skills like continuous learning and digital culture proficiency for success in AI.

It is crucial to prepare the workforce for this transition and generate new employment openings in developing industries. AI adoption is vital for successfully integrating AI tools into their methodologies. To effectively incorporate AI into the educational sector, a robust infrastructure, reassessment of academic curricula, future-oriented training for faculty and students, collaboration with relevant institutions, and integration of AI technologies are crucial (Faraj, 2022). For successful implementation of AI, practical strategies include gathering employee feedback, fostering constructive attitudes, supporting job crafting, and tailoring AI training to improve performance can be conducted (He, Teng, & Song, 2024).

The workforce must adapt by acquiring new competencies and knowledge. Collaboration between governments, educational institutions, and organizations is crucial for offering training programs and resources that prepare individuals for the age of automation, fostering creativity, critical thinking, problem-solving capabilities, and technical skills related to AI and emerging technologies. Governments and organizations should allocate resources towards research and development initiatives to generate new employment opportunities in emerging industries like AI development. By leveraging these areas and fostering entrepreneurship, a sustainable future can be established where humans and AI coexist in a balanced manner (Idrisi et al., 2024). Finally, while there are concerns about job displacement from AI automation, viewing this transition as an opportunity is crucial. By providing upskilling programs, creating new employment potentialities in developing sectors, and guaranteeing a seamless transition into a future where humans collaborate with AI technology for shared benefits (Idrisi et al., 2024).

2.5 Global Policy Responses and Frameworks Addressing the Challenges of Automation

The gradual progress of automation and AI poses significant challenges to global labour markets, economies, and societies. As a response, numerous international policy frameworks and initiatives have been established to mitigate risks and capitalize on these technologies’ benefits. The OECD AI Principles, established in 2019 and revised in May 2024, reflect recent technological advancements and policy changes, ensuring their relevance and effectiveness in the field. These principles aim to promote the development and application of AI in a reliable, innovative, and democratic manner, ensuring human rights and democratic principles are upheld. These principles consist of “inclusive growth, sustainable development, and well-being,” “human rights and democratic values, including fairness and privacy,” “transparency and explainability,” “robustness, security, and safety,” and “accountability.” Its recommendations comprise “investing in AI research and development,” “fostering a digital ecosystem for AI,” “Shaping an enabling policy environment for AI,” “building human capacity and preparing for labour market transformation,” and “international cooperation for trustworthy AI.” Nations widely adopt these principles to guide their national AI policies (OECD, 2024).

The G20 countries emphasized the significance of global collaboration on AI strategies, pledging to promote the digital economy and AI through international cooperation in policy and governance to foster international collaboration and advance discussions on AI governance to fully realize its benefits and mitigate associated risks with aims to distribute AI’s economic benefits equitably and mitigate labour market risks. The G20 AI Principles, established in 2019, aim to enhance AI solutions in the digital economy. They advocate for a regulatory framework that promotes innovation, addresses potential risks, and promotes responsible AI practices to accomplish the Sustainable Development Goals (SDGs) (G20, 2019). The United Nations Development Program (UNDP) emphasizes the importance of inclusive and sustainable economic growth in automation, aiming to align AI and automation with the UN’s SDGs, particularly in job creation, education, and reducing inequality (UNDP, 2024).

The International Labor Organization (ILO) promotes international policies to improve social protection systems and facilitate worker reskilling and upskilling, especially in at-risk sectors, in response to the effects of automation on employment, aiming to reduce displacement and maximize job opportunities (ILO, 2024). The EU AI Act is a pioneering regulation of AI. The European Union has managed automation by implementing regulatory measures such as the General Data Protection Regulation and the forthcoming AI Act. These regulations are designed to oversee the application of AI in critical sectors while fostering innovation. Additionally, the EU has prioritized reskilling efforts, offering financial support through initiatives like the Digital Europe Programme (EU, 2024). Global policy responses to automation challenges emphasize integrating innovation, economic development, social safety nets, and workforce preparedness. Countries are addressing this by allocating resources for reskilling, implementing AI regulations, and promoting international cooperation. However, challenges persist in ensuring equal benefits for all demographic groups while addressing associated risks.

Preceding studies have examined the evolution of AI and automation such as examining the effects of automation on labour demand (Aghion et al., 2022), Risk of automation for jobs in OECD countries (Arntz et al., 2016), revisiting the risk of automation (Arntz et al., 2017), skill content of recent technological change (Autor et al., 2003), influence of AI and Its challenges (Beebeejaun & Gunputh, 2023), impact of AI on the labour market (Bian, 2024), computer technology and probable job destructions (David, 2017), impact of digital work on work–life balance and job performance (Duan et al., 2023), automation technologies and their impact on employment (Filippi et al., 2023), robots at work (Graetz & Michaels, 2018), AI and unemployment in high-tech developed countries (Guliyev, 2023), dynamics of job displacement and evolution of AI (Idrisi et al., 2024), impact of automation on Canada’s workforce (Lamb, 2016), effects of digital technology on banking, production, and employment (Liu et al., 2022), technological anxiety and the future of economic growth (Mokyr et al., 2015), automation of employment in the presence of industry 4.0 (Ramos et al., 2022), and impact of AI on employment (Shen & Zhang, 2024). However, no study has been conducted to examine the evolution of job displacement in the age of AI and Automation for the last four decades. Therefore, this study was conducted to fill the literature gap.

3.1 Data Collection

The initial step involved selecting an appropriate database for the bibliometric analysis. The Scopus database was chosen due to its comprehensive coverage of peer-reviewed scientific literature and its robust citation indexing capabilities. A structured search query was carefully crafted to ensure the retrieval of relevant documents. This query combined keywords related to AI, machine learning, automation, and their impacts on job displacement and employment dynamics. The specific search string utilized was:

TITLE-ABS-KEY ((“Artificial Intelligence” OR “AI” OR “Machine Learning” OR “Automation”) AND (“Job Displacement” OR “Workforce Displacement” OR “Job Loss” OR “Employment Impact” OR “Automation Risk”) AND (industry OR sector OR field OR discipline OR department)) AND (LIMIT-TO (LANGUAGE, “English”)) AND (LIMIT-TO (DOCTYPE, “ar”) OR LIMIT-TO (DOCTYPE, “cp”) OR LIMIT-TO (DOCTYPE, “ch”) OR LIMIT-TO (DOCTYPE, “re”)) AND (EXCLUDE (PREFNAMEAUID, “Undefined”)).

This search was conducted on 9 September 2024, yielding a total of 225 documents that were deemed relevant for analysis. The selection criteria ensured that only articles published in English and classified as research articles, conference papers, or book chapters were included, providing a focused view of the literature.

3.2 Document Selection Criteria

To ensure a robust and relevant dataset, the following selection criteria were applied during the document retrieval process:

• Database Selection: Documents were sourced exclusively from the Scopus database due to its comprehensive indexing of high-quality, peer-reviewed journals across disciplines.

• Time Frame: The analysis covered the period from 1984 to 2024 to capture longitudinal trends in research on AI, automation, and job displacement.

• Language: Only documents published in English were included to ensure uniformity in content analysis and avoid potential translation inaccuracies.

• Document Type: The focus was on peer-reviewed journal articles, conference proceedings, and review articles to prioritize scholarly content with significant contributions to the field. Books, editorials, and grey literature were excluded.

• Keyword Search: A keyword search was conducted using terms such as “AI,” “automation,” “job displacement,” “workforce adaptation,” and “bibliometrics.” Boolean operators and wildcard symbols were used to refine search results and capture a broad range of relevant studies.

• Exclusion Criteria: Articles with insufficient bibliometric data, duplicate entries, or content outside the scope of AI and job displacement (e.g., focusing on unrelated industries or non-technological contexts) were excluded.

• Rationale: These criteria were designed to minimize bias, ensure data relevance, and provide a focused analysis of the most impactful research in this domain.

3.3 Selection Criteria and Limitations

The selection of documents was restricted to English-language publications indexed in the Scopus database. This approach introduces potential biases, particularly excluding non-English literature and grey literature, which may reflect diverse perspectives or emerging insights outside traditional publication channels.

3.4 Potential Biases in the Scopus Database

While the Scopus database offers comprehensive indexing of scholarly content across multiple disciplines, it is not without limitations. Language Bias is evident as Scopus primarily indexes English-language publications, potentially underrepresenting significant research published in other languages. Disciplinary Bias may occur, with certain fields like social sciences and humanities being underrepresented compared to STEM disciplines, leading to an incomplete view of interdisciplinary contributions to AI and job displacement. Regional Bias is another concern, as research outputs from developing countries or regions with limited publication infrastructure might be disproportionately excluded, skewing global representation. Additionally, Grey Literature such as policy reports, working papers, and government documents, which could provide valuable practical insights, is excluded from the database. Lastly, Recency Bias may arise due to the indexing process potentially delaying the inclusion of the most recent publications, omitting cutting-edge developments.

3.5 Mitigation Measures

To address these biases, efforts were made to diversify the dataset through the inclusion of multidisciplinary search terms, ensuring a wide range of subject areas. Future studies could complement Scopus-based analyses with data from other sources such as Web of Science, Google Scholar, or regional databases to improve representativeness and reduce bias. By incorporating these additional sources, researchers can gain a more comprehensive and balanced understanding of the impact of AI and job displacement across different disciplines, regions, and types of literature.

3.6 Data Analysis Tools

Two primary tools were utilized to analyse bibliometric data: R Studio and VOSviewer.

3.7 Data Analysis Process

The data analysis process involved several steps.

5.1 Significance of Findings

The findings of this study underscore the accelerating global interest in understanding and mitigating the impacts of AI-driven job displacement. The dominance of terms like “automation” and “ethics” reflects a broadening scope in the literature that extends beyond mere technological impacts to include social and ethical dimensions. As noted by Martini, Bellisario, and Coletti (2024), AI research is increasingly integrating human-centred concerns, recognizing that technological advancement should be managed to benefit society as a whole.

The increasing focus on ethical concerns and workforce adaptability in our study aligns with trends observed in recent literature. A shift from early concerns about job loss toward more nuanced discussions about human–AI collaboration, upskilling, and ethical AI deployment has been noted by Susskind (2020). This suggests that the academic discourse is evolving in a direction that balances technological advancement with human welfare, acknowledging the importance of proactive workforce adaptation strategies.

5.2 Policy Implications of the Findings

The findings of this bibliometric analysis carry significant implications for policymakers grappling with the challenges and opportunities posed by AI and automation in the workforce. The identification of emerging research themes, such as workforce adaptability and ethical considerations, underscores the need for proactive and adaptive policies. Policymakers should prioritize investment in large-scale reskilling and upskilling programs to mitigate the risks of job displacement while enabling workers to transition into new roles created by technological advancements.

Additionally, the global collaboration trends observed in this study suggest that international cooperation is essential for addressing the broader socio-economic impacts of automation. Governments could benefit from establishing cross-border agreements to share best practices and fund collaborative research initiatives. Ethical considerations, another key theme, call for the development of regulatory frameworks that ensure AI deployment respects privacy, equity, and human dignity.

5.3 Practical Implications

The implications of this study are both academic and practical. Policymakers need to consider the global trends identified in this study when crafting responses to the labour market disruptions caused by AI. As AI continues to evolve, reskilling initiatives and social safety nets must be prioritized to minimize the negative impacts on workers. This is consistent with policy recommendations from the OECD (2024), which emphasize the importance of education and continuous learning in an AI-driven economy.

5.4 Future Research

Future research should focus on more localized and sector-specific studies, particularly in emerging economies where the effects of automation may differ from those observed in more developed countries. Additionally, there is a need for deeper exploration of the ethical implications of AI on job displacement, particularly concerning inequality, as proposed by scholars like West (2018). More comprehensive studies could further illuminate the role of AI in job creation versus job destruction, helping to inform strategies that balance the two in the face of rapid technological advancement.

Finally, while this study highlights the global nature of AI-related job displacement research, future work should investigate whether the knowledge produced in developed countries is transferable to different socio-economic contexts. Such work could examine the scalability and applicability of workforce adaptation strategies across different industries and regions, an area highlighted by Davenport and Kirby (2016) as crucial for future research on AI’s impact.