Megatrends and the Future of Social Protection

Raphaela Hyee

Recent years have seen impressive advances in robotics and Artificial Intelligence (AI), that seemed set to transform the world of work. AI, in particular, has advanced at an astounding pace in the areas of image and speech recognition, natural language processing, translation, reading comprehension, computer programming and predictive analytics (Georgieff and Hyee, 2021[1]). This rapid progress has been accompanied by concern about the possible effects of AI and robotics deployment on the labour market, including on worker displacement.

However, employment levels have been rising in most OECD countries for decades (see Section 3.1), even as advances in robotics and information technology have enabled the automation of some tasks. While employment growth has a variety of different contributing factors, upward trends in employment rates seem at odds with fears around technological unemployment.

Indeed, empirical evidence on how past waves of technological change have impacted labour markets document positive net effects on job creation (see below). In general, technological progress improves labour efficiency by (partially) taking over/speeding up tasks performed by workers. This leads to an increase in output per effective labour input and a reduction in production costs. The employment effects of this process are ex-ante ambiguous: employment may fall as tasks are automated (substitution or displacement effect). On the other hand, lower production costs may increase output if there is sufficient demand for the good/service (productivity effect).1 New technologies may also create new tasks, leading to the emergence of new jobs (reinstatement effect). Thus, while some jobs may disappear and tasks within jobs may change, new jobs can also be created (Georgieff and Hyee, 2021[1]). Which of these effects dominates is an empirical question.

In response to a widely cited prediction that 47% of US jobs were at risk of being automated over the next 10‑20 years (Frey and Osborne, 2017[2]), the OECD created an alternative measure of automation risk by occupation that placed greater emphasis on the heterogeneity of task-content within occupations (Nedelkoska and Quintini, 2018[3]). This approach argues that an occupation consists of a bundle of tasks requiring different skills that are performed together, some of which continue to be difficult or impossible to automate, e.g. social intelligence (the ability to effectively negotiate complex social relationships) and perception and manipulation (the ability to carry out physical tasks in an unstructured work environment). This approach resulted in a much-reduced estimate of jobs at high risk of automation: an average of 14% across countries, and 10% for the United States in 2012.

Georgieff and Milanez (2021[4]) look at whether countries and jobs that were deemed to be at high risk of automation back in 2012 experienced employment declines over the subsequent decade using data from labour force surveys from 21 countries at the occupation level (38 occupations) over the time period 2012 to 2019. The results show no support for net job destruction at the broad country level. All countries in the study saw employment growth over the observation period (2012 to 2019), and countries that faced higher overall automation risk in 2012 did not experience lower employment growth until 2019. In fact, countries where occupations faced higher automation risk back in 2012 experienced higher occupational employment growth over the subsequent period. This is consistent with a story in which automation contributes to positive employment growth through productivity growth: increases in labour productivity lead to lower prices on consumer goods; and lower prices boost consumer demand, which in turn increases employment levels (even if the amount of labour per unit has declined).

Within countries, the majority of occupations also saw employment growth. However, occupations that were at higher risk of automation experienced lower employment growth or even modest declines in employment levels. On average across countries, employment among the riskiest half of occupations grew by 6% compared to 18% among the least risky. Thus, there is evidence that automation has worsened employment prospects for some workers. The occupations that saw employment declines include: skilled agricultural workers; clerical support workers; skilled forestry, fishing and hunting workers; handicraft and printing workers; and metal and machinery workers. These declines are even more striking given that they occur against a backdrop of rising employment across countries.

Doorley et al. (2023[5]) similarly find only small negative effects of robot penetration in 14 European countries between 2006 and 2018. They also look at the effects on wages, and find stronger effects, with wage losses concentrated in the bottom half of the wage distribution. Income support programmes cushioned this negative effect, especially in Western European countries.

There are reasons to believe that its impact on employment may be different from previous waves of technological progress. Autor, Levy and Murnane (2003[6]) postulated that jobs consist of routine (and thus in principle programmable) and non-routine tasks. Previous waves of technological progress were primarily associated with the automation of routine tasks. Computers, for example, are capable of performing routine cognitive tasks including record-keeping, calculation, and searching for information. Similarly, industrial robots are programmable manipulators of physical objects and therefore associated with the automation of routine manual tasks such as welding, painting or packaging (Raj and Seamans, 2019[7]).2 These technologies therefore mainly substitute for workers in low- and middle‑skill occupations.

Tasks typically associated with high-skilled occupations, such as non-routine manual tasks (requiring dexterity) and non-routine cognitive tasks (requiring abstract reasoning, creativity, and social intelligence) were previously thought to be outside the scope of automation (Autor, Levy and Murnane, 2003[6]; Acemoglu and Restrepo, 2020[8]).

However, recent advances in AI mean that non-routine cognitive tasks can also increasingly be automated (Lane and Saint-Martin, 2021[9]). In most of its current applications, AI refers to computer software that relies on highly sophisticated algorithmic techniques to find patterns in data and make predictions about the future. Analysis of patent texts suggests AI is capable of formulating medical prognosis and suggesting treatment, detecting cancer and identifying fraud (Webb, 2020[10]). Thus, in contrast to previous waves of automation, AI might disproportionally affect high-skilled workers. Importantly, AI can learn from its actions, and improve its predictions and recommendations over time (OECD, 2023[11]).

Georgieff and Hyee (2021[1]) look at the links between AI and employment growth in a cross-country context. They adapt the AI occupational impact measure proposed by Felten, Raj and Seamans (2018[12]; 2019[13]) – an indicator measuring the degree to which occupations rely on abilities in which AI has made the most progress in recent years – and extend it to 23 OECD countries by linking it to the Survey of Adult Skills, PIAAC. This indicator, which allows for variations in AI exposure across occupations, as well as within occupations and across countries, is matched to Labour Force Surveys to analyse the relationship with employment growth. It is important to point out, however, that their measure pre‑dates the most recent advances in generative AI applications (ChatGPT for example) – both the occupational range and extent of AI exposure might rapidly become larger as generative AI use is increasingly incorporated into production processes and new, more powerful AI systems are developed.

Georgieff and Hyee (2021[1]) find that business professionals, managers, chief executives and science and engineering professionals are most exposed to AI. This follows intuitively from recent AI progress in non-routine, cognitive tasks. In contrast, “Food preparation assistants,” “Agriculture, forestry and fishery labourers” and “Cleaners, helpers” are the least exposed to AI.

Georgieff and Hyee (2021[1]) find that, over the period 2012‑19, there is no clear relationship between AI exposure and employment growth across all occupations. Moreover, in occupations where computer use is high, AI appears to be positively associated with employment growth. There is also some evidence of a negative relationship between AI exposure and growth in average hours worked among occupations where computer use is low. While further research is needed to identify the exact mechanisms driving these results, one possible explanation is that partial automation by AI increases productivity directly as well as by shifting the task composition of occupations towards higher value‑added tasks. This increase in labour productivity and output counteracts the direct displacement effect of automation through AI for workers with good digital skills, who may find it easier to use AI effectively and shift to non-automatable, higher-value tasks within their occupations. The opposite could be true for workers with poor digital skills, who may be unable to interact efficiently with AI and thus miss all potential benefits of the technology.

Felten, Raj and Seamans (2019[13]) look at the effect of exposure to AI on employment and wages in the United States at the occupational level. They do not find any link between AI exposure and (aggregate) employment, but they do find a positive effect of AI exposure on wage growth, suggesting that the productivity effect of AI may outweigh the substitution effect. This effect on wage growth is concentrated in occupations that require software skills and in high-wage occupations.

Even at lower levels of aggregation, there is currently no detectable effect of AI on aggregate employment. Acemoglu et al. (2022[14]) examine employment changes from AI using differences in AI exposure by US commuting zones by industry, and separately, variation by detailed occupation. In all specifications they find no statistically significant effect of AI exposure on employment between 2010 and 2017 (industry) and 2018 (occupation). Fossen and Sorgner (2022[15]) estimate the effect of AI exposure on the probability of leaving employment using short panels of individual workers in the United States between 2011 and 2018. They find that exposure to AI decreases the probability of workers to exit employment.

Surveys of firms’ AI adoption and ensuing employment changes similarly find no detectable decreases in employment. In a survey of 759 managers of firms in the United Kingdom, Hunt, Sarkar and Warhurst (2022[16]) find that AI is leading to greater turnover, but when the researchers look at net employment changes, the results are inconclusive as firms are equally likely to report net employment gains and losses compared to firms not adopting AI. Similarly, an OECD survey of firms in manufacturing and finance across seven OECD countries finds that most firms adopting AI say that it did not change employment (Lane, Williams and Broecke, 2023[17]). Slightly more firms report employment decreases compared to increases, but these differences are not statistically significant. These results accord with national surveys of firms that adopt AI. A random sample of over 300 000 employer businesses in the United States from the U.S. Census Bureau collected information on firms’ adoption of advanced technologies from 2016 to 2018. The majority of firms reported no changes in employment levels due to advanced technologies but, of the firms adopting AI, 26% said that this caused them to increase employment compared to less than 10% which saw their employment levels decrease (Acemoglu et al., 2022[18]).

Much like the effect of new technologies on employment, the effects on wages are also ex-ante indetermined: if the substitution effect dominates, AI could exert downward pressure on wages, or it could increase wages through a productivity effect. To harness this productivity effect, workers need to both learn to work effectively with the new technology and to adapt to the changing task composition that puts more emphasis on tasks that AI cannot yet perform. Such adaptation may be costly and will depend on worker characteristics (e.g. education, familiarity with other digital technologies, etc. (Georgieff, 2024[19]).

While high-skilled workers may be more exposed to recent advances in AI, they also rely more than other workers on high value‑added abilities for which AI is not yet very advanced, such as inductive reasoning, creative or social intelligence (Lane and Saint-Martin, 2021[9]; Lassébie and Quintini, 2022[20]). Moreover, high-skilled workers often find it easier to adapt to new technologies because they are more likely to already work with digital technologies and participate more in training, which puts them in a better position than lower-skilled workers to reap the potential benefits of AI (Georgieff and Hyee, 2021[1]).

To date, most empirical studies on the wage effects of automation technology have focused on industrial robots, and most of these studies suggest that substitution effects prevail (Georgieff, 2024[19]). Acemoglu and Restrepo (2020[21]) find a negative relationship between the adoption of industrial robots and wages across US commuting zones. Dauth et al. (2017[22]) extend this analysis to Germany and find similar results for medium-skilled workers in machine operating occupations. Webb (2020[10]) finds that US occupations heavily exposed to industrial robots have experienced wage declines.

The emerging evidence on the impact of AI on wage inequality so far is mixed (Georgieff, 2024[19]). Two studies show that exposure to AI is positively associated with wage growth at the occupation level (Felten, Raj and Seamans, 2019[13]) and the individual level (Fossen and Sorgner, 2019[23]), particularly among those with higher wages and/or higher levels of education. However, Acemoglu et al. (2020[24]) find no relationship between exposure to AI and wage growth at the occupation or industry level.

Automation does not only affect inequalities between occupations, but can also affect inequalities within occupations. For example, van der Velde (2020[25]) shows that computerisation has increased differences in performance between workers in a given occupation. This could be due to a combination of two effects: i) computerisation reduces the prevalence of routine tasks in an occupation; and ii) differences in skill levels among workers performing more routine tasks have a lower impact on output (Jung and Mercenier, 2014[26]) – notably because routine tasks leave workers with little autonomy (Oldenski, 2012[27]; Marcolin, Miroudot and Squicciarini, 2016[28]) and few opportunities to exploit their creativity (Frey and Osborne, 2017[2]).

By contrast, there is evidence that the use of (generative) AI can reduce differences in performance between workers within an occupation (Brynjolfsson, Li and Raymond, 2023[29]; Choi and Schwarcz, 2023[30]; Dell’Acqua et al., 2023[31]; Haslberger, Gingrich and Bhatia, 2023[32]; Noy and Zhang, 2023[33]; Peng et al., 2023[34]), which could diminish wage inequality. This could be explained by the fact that AI systems are trained to predict good outcomes, and will therefore embody the practices of high performers (Brynjolfsson, Li and Raymond, 2023[29]). Low performers therefore have more to gain from using AI. AI can also reduce performance differences within an occupation through a selection effect, if low-performing workers leave their job because they are unable to adapt to AI tools by shifting their activities to tasks that AI cannot automate. For example, some stock analysts who were unable to adapt to AI-based prediction tools by shifting their work to more social activities may have left the profession (Grennan and Michaely, 2017[35]).

Georgieff (2024[19]) looks at the link between AI and wage inequality in a cross-country context. It uses the cross-country measure of occupational exposure to AI derived by Georgieff and Hyee (2021[1]) from that developed by Felten, Raj and Seamans (2019[13]). He matches this measure to the Structure of Earnings Survey (SES) and US Current Population Survey (CPS) data to analyse the relationship with wage inequality. He finds that there is no indication that AI has affected wage inequality between occupations over the period 2014‑18. However, AI may be lowering wage inequality within occupations – consistent with the above‑mentioned literature that AI might be performance enhancing for lower-productivity workers. However, this does not appear to affect gender or age wage gaps: within occupations, exposure to AI is not associated with changes in wage inequality between demographic groups.

However, it is important to note that the above quoted empirical evidence is necessarily backwards looking, using data, although recent, from a time when AI adoption was still relatively low. Given the speed with which AI and its adoption in the economy progresses, these results should be interpreted with caution.

Results from the OECD’s 2022 Risks that Matter Survey show that, across 27 countries, people generally believe that technology will help them in their jobs, with 34‑40% of people, on average, believing that technology will either a) make their working hours more compatible with their private life, b) make their job less dangerous or physically demanding, or c) make their jobs less boring, repetitive, stressful or mentally demanding. This compares to an average of about 18‑21% of people who believe that they will likely lose their job due to technology.

These sentiments did, however, vary considerably across the education and income distribution. Those with higher levels of education (i.e. tertiary) were significantly more likely to believe that technology would have a positive impact on their job. It is noteworthy that people with higher education were also more likely to believe that their job would be replaced by technology, although these gaps were much smaller.

To date (and this evidence is necessarily backwards looking), worries about widespread technology driven job displacement have not been borne out, and the evidence on the effects of AI on wages, though mixed, seems to be indicating that if there is an effect, it seems to be positive, and attenuating inequality.

In the absence of population growth, rising productivity through technological progress is the only way to achieve long-term economic growth, which is necessary for the continued funding of public expenditure in a non-zero interest rate environment (Arslanalp and Eichengreen, 2023[36]). Together with higher labour supply (through longer working lives and tapping the full potential of women and part-time workers, see Chapters 2 and 3), increased productivity through technological progress may therefore be necessary for the sustainability of social protection systems.

Public policies can help to ensure that workers benefit from the productivity gains and cost savings generated by AI, which will also support social protection funding that in many countries mostly relies on labour income. For instance, worker skills will be a crucial determinant to the extent that new technologies including AI increases worker productivity and wages. Governments should support training within companies as well as through formal education institutions including life‑long learning. Public policy can also support collective bargaining and a fair distribution of productivity gains within firms (OECD, 2023[11]).

Countries may also want to reconsider their tax mixes should a higher share of the technology-driven productivity growth accrue to capital. To date, evidence on whether the labour share of income is falling is mixed (see, e.g. (Gutiérrez and Piton, 2020[37]; Bergholt, Furlanetto and Maffei-Faccioli, 2022[38])), but increased automatisation certainly has the potential to drive down the share of GDP claimed by workers. Rebalancing labour and capital taxation may therefore be necessary to protect social protection funding if productivity gains do not translate into higher employment and wages.

The relative taxation of capital and labour may also directly influence AI adoption. Many countries apply a lower effective tax rate on capital than on labour. Social security contributions, which are typically only levied on labour income, contribute to this imbalance. Where effective taxes on labour are higher than on capital, firms can be pushed to automate above the level they would in the absence of this implicit tax subsidy (OECD, 2023[11]).

AI could also be used to improve the administration of social protection systems, e.g. by automating tasks such as data entry and document processing, saving staff time and lowering costs. Some countries are using AI for fraud detection in benefit claims, improving the targeting of benefits. AI may also increase the accessibility of social benefits, e.g. through chatbots that can answer personalised questions and assist claimants through the claims process (OECD, 2024[39]).

[18] Acemoglu, D. et al. (2022), Automation and the Workforce: A Firm-Level View from the 2019 Annual Business Survey, National Bureau of Economic Research, Cambridge, MA, https://doi.org/10.3386/w30659.

[14] Acemoglu, D. et al. (2022), “Artificial Intelligence and Jobs: Evidence from Online Vacancies”, Journal of Labor Economics, Vol. 40/S1, pp. S293-S340, https://doi.org/10.1086/718327.

[24] Acemoglu, D. et al. (2020), AI and Jobs: Evidence from Online Vacancies, National Bureau of Economic Research, Cambridge, MA, https://doi.org/10.3386/w28257.

[8] Acemoglu, D. and P. Restrepo (2020), “Robots and jobs: Evidence from us labor markets”, Journal of Political Economy, Vol. 128/6, pp. 2188-2244, https://www.journals.uchicago.edu/doi/abs/10.1086/705716.

[21] Acemoglu, D. and P. Restrepo (2020), “Robots and Jobs: Evidence from US Labor Markets”, Journal of Political Economy, Vol. 128/6, pp. 2188-2244, https://doi.org/10.1086/705716.

[36] Arslanalp, S. and B. Eichengreen (2023), “Living with High Public Debt”, Paper presented at the 2023 Jackson Hole Economic Symposium, https://www.kansascityfed.org/research/jackson-hole-economic-symposium/economic-symposium-conference-proceedings/.

[6] Autor, D., F. Levy and R. Murnane (2003), “The Skill Content of Recent Technological Change: An Empirical Exploration”, The Quarterly Journal of Economics, Vol. 118/4, pp. 1279-1333, https://academic.oup.com/qje/article-lookup/doi/10.1162/003355303322552801.

[38] Bergholt, D., F. Furlanetto and N. Maffei-Faccioli (2022), “The Decline of the Labor Share: New Empirical Evidence”, American Economic Journal: Macroeconomics, Vol. 14/3, pp. 163-198, https://doi.org/10.1257/mac.20190365.

[40] Bessen, J. (2015), “How Computer Automation Affects Occupations: Technology, Jobs, and Skills”, SSRN Electronic Journal, https://doi.org/10.2139/ssrn.2690435.

[29] Brynjolfsson, E., D. Li and L. Raymond (2023), Generative AI at Work, National Bureau of Economic Research, Cambridge, MA, https://doi.org/10.3386/w31161.

[30] Choi, J. and D. Schwarcz (2023), “AI Assistance in Legal Analysis: An Empirical Study”, SSRN Electronic Journal, https://doi.org/10.2139/ssrn.4539836.

[22] Dauth, W. et al. (2017), German Robots—The Impact of Industrial Robots on Workers.

[31] Dell’Acqua, F. et al. (2023), “Navigating the Jagged Technological Frontier: Field Experimental Evidence of the Effects of AI on Knowledge Worker Productivity and Quality”, SSRN Electronic Journal, https://doi.org/10.2139/ssrn.4573321.

[5] Doorley, K. et al. (2023), “Automation and Income Inequality in Europe”, IZA Discussion Paper No 16499, https://ssrn.com/abstract=4595178.

[13] Felten, E., M. Raj and R. Seamans (2019), “The Occupational Impact of Artificial Intelligence: Labor, Skills, and Polarization”, NYU Stern School of Business, https://doi.org/10.2139/ssrn.3368605.

[12] Felten, E., M. Raj and R. Seamans (2018), “A Method to Link Advances in Artificial Intelligence to Occupational Abilities”, AEA Papers and Proceedings, Vol. 108, pp. 54-57.

[15] Fossen, F. and A. Sorgner (2022), “New digital technologies and heterogeneous wage and employment dynamics in the United States: Evidence from individual-level data”, Technological Forecasting and Social Change, Vol. 175, p. 121381, https://doi.org/10.1016/j.techfore.2021.121381.

[23] Fossen, F. and A. Sorgner (2019), “New Digital Technologies and Heterogeneous Employment and Wage Dynamics in the United States: Evidence from Individual-Level Data”, SSRN Electronic Journal, https://doi.org/10.2139/ssrn.3390231.

[2] Frey, C. and M. Osborne (2017), “The future of employment: How susceptible are jobs to computerisation?”, Technological Forecasting and Social Change, Vol. 114, pp. 254-280, https://doi.org/10.1016/j.techfore.2016.08.019.

[19] Georgieff, A. (2024), “Artificial Intelligence and Wage Inequality”, OECD Artificial Intelligence Papers, No. 13, OECD Publishing, Paris, https://doi.org/10.1787/bf98a45c-en.

[1] Georgieff, A. and R. Hyee (2021), “Artificial intelligence and employment : New cross-country evidence”, OECD Social, Employment and Migration Working Papers, No. 265, OECD Publishing, Paris, https://doi.org/10.1787/c2c1d276-en.

[4] Georgieff, A. and A. Milanez (2021), “What happened to jobs at high risk of automation?”, OECD Social, Employment and Migration Working Papers, No. 255, OECD Publishing, Paris, https://doi.org/10.1787/10bc97f4-en.

[35] Grennan, J. and R. Michaely (2017), “Artificial Intelligence and the Future of Work: Evidence from Analysts”, NBER Economics of Artificial Intelligence Conference Paper, https://conference.nber.org/conf_papers/f130049.pdf (accessed on 29 May 2020).

[37] Gutiérrez, G. and S. Piton (2020), “Revisiting the Global Decline of the (Non-housing) Labor Share”, American Economic Review: Insights, Vol. 2/3, pp. 321-338, https://doi.org/10.1257/aeri.20190285.

[32] Haslberger, M., J. Gingrich and J. Bhatia (2023), “No Great Equalizer: Experimental Evidence on AI in the UK Labor Market”, SSRN Electronic Journal, https://doi.org/10.2139/ssrn.4594466.

[16] Hunt, W., S. Sarkar and C. Warhurst (2022), “Measuring the impact of AI on jobs at the organization level: Lessons from a survey of UK business leaders”, Research Policy, Vol. 51/2, p. 104425, https://doi.org/10.1016/j.respol.2021.104425.

[26] Jung, J. and J. Mercenier (2014), “Routinization‐biased technical change and globalization: understanding labor market polarization”, Economic Inquiry, Vol. 52/4, pp. 1446-1465, https://doi.org/10.1111/ecin.12108.

[9] Lane, M. and A. Saint-Martin (2021), “The impact of Artificial Intelligence on the labour market: What do we know so far?”, OECD Social, Employment and Migration Working Papers, No. 256, OECD Publishing, Paris, https://doi.org/10.1787/7c895724-en.

[17] Lane, M., M. Williams and S. Broecke (2023), “The impact of AI on the workplace: Main findings from the OECD AI surveys of employers and workers”, OECD Social, Employment and Migration Working Papers, No. 288, OECD Publishing, Paris, https://doi.org/10.1787/ea0a0fe1-en.

[20] Lassébie, J. and G. Quintini (2022), “What skills and abilities can automation technologies replicate and what does it mean for workers?: New evidence”, OECD Social, Employment and Migration Working Papers, No. 282, OECD Publishing, Paris, https://doi.org/10.1787/646aad77-en.

[28] Marcolin, L., S. Miroudot and M. Squicciarini (2016), “GVCs, Jobs And Routine Content Of Occupations”, OECD Trade Policy Papers, No. 187, OECD Publishing, Paris, https://doi.org/10.1787/5jm0mq7kr6s8-en.

[3] Nedelkoska, L. and G. Quintini (2018), “Automation, skills use and training”, OECD Social, Employment and Migration Working Papers, No. 202, OECD Publishing, Paris, https://doi.org/10.1787/2e2f4eea-en.

[41] Nolan, A. (2021), “Making life richer, easier and healthier: Robots, their future and the roles for public policy”, OECD Science, Technology and Industry Policy Papers, No. 117, OECD Publishing, Paris, https://doi.org/10.1787/5ea15d01-en.

[33] Noy, S. and W. Zhang (2023), “Experimental Evidence on the Productivity Effects of Generative Artificial Intelligence”, SSRN Electronic Journal, https://doi.org/10.2139/ssrn.4375283.

[39] OECD (2024), Modernising Access to Social Protection: Strategies, Technologies and Data Advances in OECD Countries, OECD Publishing, Paris, https://doi.org/10.1787/af31746d-en.

[11] OECD (2023), OECD Employment Outlook 2023: Artificial Intelligence and the Labour Market, OECD Publishing, Paris, https://doi.org/10.1787/08785bba-en.

[27] Oldenski, L. (2012), “The task composition of offshoring by U.S. multinationals”, Économie internationale, Vol. n° 131/3, pp. 5-21, https://doi.org/10.3917/ecoi.131.0005.

[34] Peng, S. et al. (2023), “The Impact of AI on Developer Productivity: Evidence from GitHub Copilot”.

[7] Raj, M. and R. Seamans (2019), “Primer on artificial intelligence and robotics”, Journal of Organisation Design, Vol. 8/11, https://doi.org/10.1186/s41469-019-0050-0.

[25] van der Velde, L. (2020), “Within Occupation Wage Dispersion and the Task Content of Jobs”, Oxford Bulletin of Economics and Statistics, Vol. 82/5, pp. 1161-1197, https://doi.org/10.1111/obes.12368.

[10] Webb, M. (2020), “The Impact of Artificial Intelligence on the Labor Market”, Working Paper, Standford University, https://dx.doi.org/10.2139/ssrn.3482150.