Are Our Jobs at Risk?
Estimating the Effect of Artificial Intelligence on the Swiss Labor Market
Article Summary
Part 1: Introduction to AI and a task-based view on occupations
Note: This blog post summarizes the article “Are Our Jobs at Risk? Estimating the Effect of Artificial Intelligence on the Swiss Labor Market”. Authors: Timon Jaeggi, Benjamin Schaefer, Christian Dietzmann, Reinhard Jung, and Ulrich Matter
Available: https://journals.aom.org/doi/abs/10.5465/AMPROC.2023.11166abstract
Suggested citation: Jaeggi, T., Schaefer, B., Dietzmann, C., Jung, R., & Matter, U. (2023). Are Our Jobs at Risk? Estimating the Effect of Artificial Intelligence on the Swiss Labor Market. In Academy of Management Proceedings (Vol. 2023, No. 1, p. 11166).
Introduction
Recent advances in artificial intelligence (AI) have radically expanded the range of technical applications and our idea of the areas in which it can be used (Jaeggi et al., 2023). Until now, the prevailing notion was that humans would retain their comparative advantage, especially in creative and cognitive fields of work, as historically, rule-based computers exclusively replaced occupations with a limited and clearly defined range of tasks (Frey & Osborne, 2013). Complex tasks such as communicating with customers, developing management strategies, or driving a car are subject to irregular and dynamic rules that have long been considered impossible to imitate by computer systems (Tolan, Pesole, Marinez-Plumed, & Fernandes-Macias, 2021). However, current developments in AI represent a fundamental paradigm shift (Tolan et al., 2021; Autor, 2014). Due to their self-learning ability, machines can now perform classification tasks, predict future developments, or communicate in natural language (Brynjolfsson, Rock, & Syverson, 2019). As a result, the range of professions directly affected by AI capabilities is continuously expanding, and such developments have far-reaching consequences for organizations, their management, and the economy (Jaeggi et al., 2023). An increasing proportion of employees find themselves in direct competition with intelligent solutions that are capable of redesigning specific tasks and processes or even changing entire business models (Alt, Beck, & Smits, 2018; MacCrory, Westerman, Alhammadi, & Brynjolfsson, 2014).
Artificial intelligence
AI describes the ability of computers to perceive, think, and act (Winston, 1984). Therefore, AI is not seen as a single technology but rather as a continuous development of computational capabilities that have evolved rapidly over the past decades (McCorduck & Cfe, 2004). Today’s developments in AI are driven by advances in machine learning (ML) (Taddy, 2019). ML is considered an essential component of AI as it is a collective term for various sophisticated statistical techniques that enable the recognition of patterns from large amounts of data to derive predictions or estimates (Webb, 2019). Scientists generally consider ML as a general-purpose-technology (GPT), which means that ML can find broad applications in various fields and professions and promote the development of complementary inventions (Lane & Saint-Martin, 2021; Taddy, 2019; Brynjolfsson, Mitchell, & Rock, 2018; Bresnahan & Trajtenberg, 1995).
AI can be characterized by pure software (“software AI”) or by a physical basis (“embodied AI”) (Jaeggi et al., 2023). Software AI focuses primarily on non-routine cognitive tasks (Jaeggi et al., 2023). For example, AI assists pathologists in medical practice and learns from thousands of previously viewed scans to recognize critical features that help detect breast cancer without explicit human instruction (Jaeggi et al., 2023). In the context of creative tasks, AI can, for example, help to create new fragrances for the perfume industry (Goodwin et al., 2017; IBM Research, 2018). Unlike software AI, embodied AI systems usually work with unstructured sensory data such as spoken language, sounds, or real-time videos (Jaeggi et al., 2023). In a car factory, for example, intelligent articulated arms assemble car parts, with various sensors helping the system recognize individual objects (Jaeggi et al., 2023). In other application areas, such as education or entertainment, embodied AI has human-like characteristics that enable users to interact intuitively with the system. For example, these humanoid characteristics were realized by the robots Pepper (e.g., Pandey & Gelin, 2018) and Sophia by mapping human faces with facial expressions or gestures.
Artificial intelligence and the labor market
As a result of recent developments in AI, the impact of automation on the labor market has begun to be reassessed (Jaeggi et al., 2023). Non-routine cognitive or manual jobs, long considered unaffected by previous technologies, are increasingly threatened by self-learning algorithms, leading many economists to argue that this time, the impact on the labor market is fundamentally different from other technologies (Furman, 2019; Georgieff & Hyee, 2021). Nevertheless, it is difficult for economists to assess the consequences for occupations and the labor market as the range of software and embodied AI capabilities grows (Webb, 2019; Acemoglu et al., 2020). A key finding recognized by researchers is that technologies fundamentally impact at the task level rather than the job level (Huang & Rust, 2018; Sampson, 2020; Levy & Murnane, 1996; Acemoglu & Restrepo, 2018). For example, AI may not impact the entire profession, but only individual tasks within a profession (Sampson, 2021). A task includes various activities, varying wildly depending on the occupation (Jaeggi et al., 2023). They range from simple mechanical tasks to demanding cognitive tasks, sometimes performed within the same occupation (Huang & Rust, 2018).
A task-based view on occupations
AI’s social and economic consequences can be so different that some experts believe we are heading towards a future without jobs (Ford, 2015). In contrast, others see AI as a driving force for increasing labor productivity and improving working conditions (Damioli, Van Roy, & Vertesy, 2021). Although these views largely differ, researchers argue that both perspectives can exist in parallel (Jaeggi et al., 2023). This is because AI can replace and complement employees in their work (Frank et al., 2019). For example, Frank et al. (2019) and Seamans and Raj (2018) found that, depending on the type of activity, employees are complemented by the technology (complementation) or compete with it (substitution). As a result, supporting employees’ tasks with AI positively impacts their performance and, thus, the company’s results (Jaeggi et al., 2023). In addition, the substitution of tasks by AI can have positive effects on organizations (e.g., cost advantages or efficiency gains) and individuals through indirect performance improvements because of the potential (re)assignment of additional tasks (Jaeggi et al., 2023). However, employees quickly lose importance and are replaced by AI if they are not (re)assigned additional tasks (Jaeggi et al., 2023).
Table 1 shows how AI can affect the tasks of an occupation. It is essential to understand that an employee can simultaneously be exposed to these three types of exposure. While some tasks in an occupation may be replaced, others may be augmented by current technologies or remain unaffected (Jaeggi et al., 2023).
Table 1: Types of AI exposures
| AI task substitution | AI task completion | AI task independence |
| “Race against the machine” | “Race against the machine” | “Running a different race” |
| Complete substitution/replacement of human labor in the execution of the task. The human is no longer involved in the execution of the task. | Supplementing human work with AI, whereby the human is always the central component in the execution of the task. Types of augmentation: SensesMeasures/ActionPerception/Recognition | The task within the profession remains largely unaffected by AI. |
| Note: The table is based on Jaeggi et al. (2023) | ||
On the one hand, new technologies can take over skills and abilities (AI task substitution) previously reserved for humans, which puts employees at a relative disadvantage in their work (MacCrory et al., 2014). One example is industrialization, which replaced many workers during the Industrial Revolution. An intense match between task and technology can lead to substitutability effects that make workers redundant. On the other hand, technological innovations can complement human skills in performing their work (AI task completion) (MacCrory et al., 2014). Skills can be enhanced through collaboration between workers and digital technologies on a given task, ultimately leading to higher labor productivity (MacCrory et al., 2014). Finally, jobs may require skills and abilities that still need to be adequately covered by the capabilities of today’s technologies, leaving the employee largely unaffected (AI task independence) (Jaeggi et al., 2023). For example, professions with a high degree of personal interaction (e.g., healthcare workers, teachers, lawyers, and pastoral workers) and professions that require much creativity (e.g., sculptors or painters) have hardly changed in the early days of the computer age (Jaeggi et al., 2023).
The continuous development of computational capabilities characterized by software AI and embodied AI thus have multiple applications and different impacts on individual tasks of an occupation. In this context, it is essential to consider the task-based perspective, as the exposure of an occupation to AI can be determined by comparing the associated tasks (Jaeggi et al., 2023). In the theoretical context, the “task-technology fit” theory (TFF) is also of utmost importance (cf. Goodhue & Thompson, 1995). However, what impact will AI have on the Swiss economic sector? In particular, on the professions and employees in individual sectors? Moreover, how will it affect the labor market as a whole? What does the changed situation mean for companies and politics? These questions will be examined in the second part of this blog series, along with the presentation of results.
Literature
Acemoglu, D., & Restrepo, P. 2018. The Race between Man and Machine: Implications of Technology for Growth, Factor Shares, and Employment. American Economic Review, 108(6), 1488-1542.
Acemoglu, D., Autor, D., Hazell, J., & Restrepo, P. 2020. AI and Jobs: Evidence from Online Vacancies. Working paper no. w28257, NBER, Cambridge, Massachusetts.
Alt, R., Beck, R., & Smits, M. T. 2018. FinTech and the transformation of the financial industry. Electronic Markets, 28: 235-243.
Author, D. H. 2014. Polanyi’s Paradox and the Shape of Employment Growth. Working paper no. 20485, NBER, Cambridge, Massachusetts.
Bresnahan, T. F., & Trajtenberg, M. 1995. General purpose technologies ‘Engines of growth’? Journal of Econometrics, 65(1): 83-108.
Brynjolfsson, E., & Mitchell, T. 2017. What Can Machine Learning Do? Workforce Implications. Science, 358(6370): 1530-1534.
Brynjolfsson, E., Mitchell, T., & Rock, D. 2018. What Can Machines Learn, and What Does It Mean for Occupations and the Economy? AEA Papers and Proceedings, 108: 43-47.
Brynjolfsson, E., Rock, D., & Syverson, C. 2019. Artificial Intelligence and the Modern Productivity Paradox: A Clash of Expectations and Statistics. In A. Agrawal, J. Gans, & A. Goldfarb, The Economics of Artificial Intelligence: An Agenda, 23-57. Chicago: Uni-versity of Chicago Press.
Damioli, G., Van Roy, V., & Vertesy, D. 2021 The Impact of Artificial Intelligence on Labor Productivity. Eurasian Business Review, 11: 1-25.
Ford, M. 2015. Rise of the Robots: Technology and the Threat of a Jobless Future. New York, NY: Basic Books.
Frank, M. R., Autor, D., Bessen, J. E., Brynjolfsson, E., Cebrian, M., Deming, D. J., Rahwan, I. 2019. Toward Understanding the Impact of Artificial Intelligence on Labor. Proceedings of the National Academy of Sciences, 116(14): 6531-6539.
Frey, C. B., & Osborne, M. A. 2013. The Future of Employment: How Susceptible are Jobs to Computerization? Technological forecasting and social change, 114: 254-280.
Furman, J. 2019. Should We Be Reassured If Automation in the Future Looks Like Automation in the Past? In A. Agrawal, J. Gans, & A. Goldfarb, The Economics of Artificial Intelligence: An Agenda: 317-328. Chicago and London: University of Chicago Press.
Georgieff, A., & Hyee, R. 2021. Artificial Intelligence and Employment: New Cross-Country Evidence. OECD Social, Employment and Migration. https://www.oecd-ilibrary.org/content/paper/c2c1d276-en.
Goodwin, R., Maria, J., Das, P., Horesh, R., Segal, R., Fu, J., & Harris, C. 2017. AI for Fragrance Design. 31st Conference on Neural Information Processing Systems (NIPS 2017). Long Beach, CA: Curran Associates Inc.
Huang, M.-H., & Rust, R. T. 2018. Artificial Intelligence in Service. Journal of Service Research, 21(2): 155-172.
IBM Research 2018. Using AI to Create New Fragrances, IBM Research Blog. https://www.ibm.com/blogs/research/2018/10/ai-fragrances/.
Jaeggi, T., Schaefer, B., Dietzmann, C., Jung, R., & Matter, U. (2023). Are Our Jobs at Risk? Estimating the Effect of Artificial Intelligence on the Swiss Labor Market. In Academy of Management Proceedings (Vol. 2023, No. 1, p. 11166).
Lane, M., & Saint-Martin, A. 2021. The Impact of Artificial Intelligence on the Labor Market: What do We Know so Far? OECD Social, Employment and Migration Working Papers, no. 256, OECD Publishing, Paris.
Levy, F., & Murnane, R. J. 1996. With What Skills Are Computers a Complement? The American Economic Review, 86(2): 258-262.
MacCrory, F., Westerman, G., Alhammadi, Y., & Brynjolfsson, E. 2014. Racing With and Against the Machine : Changes in Occupational Skill Composition in an Era of Rapid Technological Advance. 35th International Conference on Information Systems. Auckland.
McCorduck, P., & Cfe, C. 2004. machines who think: A personal inquiry into the history and prospects of artificial intelligence. CRC Press.
Moore, T., & Bokelberg, E. 2019. How IBM Incorporates Artificial Intelligence into Strategic Workforce Planning. https://www.shrm.org/executive/resources/people-strategy-journal/Fall2019/Pages/moore-bokelberg-feature.aspx.
Sampson, S. 2020. Predicting Automation of Professional Jobs in Healthcare. Proceedings of the 53rd Hawaii International Conference on System Sciences, Maui, HI.
Sampson, S. 2021. A Strategic Framework for Task Automation in Professional Services. Journal of Service Research, 24(1): 122-140.
Seamans, R., & Raj, M. 2018. AI, Labor, Productivity and the Need For Firm-Level Data. Working Paper no. 24239, National Bureau of Economic Research, Cambridge, Massachusetts.
Taddy, M. 2019. The Technological Elements of Artificial Intelligence. In A. Agrawal, J. Gans, & A. Goldfarb, The Economics of Artificial Intelligence: An Agenda: 61-87. Chicago and London: University of Chicago Press.
Tolan, S., Pesole, A., Marinez-Plumed, & Fernandes-Macias, E. 2021. Measuring the Occupational Impact of AI: Tasks, Cognitive Abilities and AI Benchmarks. Journal of Artificial Intelligence Research, 71: 191-236.
Webb, M. 2019. The Impact of Artificial Intelligence on the Labor Market. SSRN.
Winston, P. H. 1984. artificial intelligence. MA, Boston: Addison-Wesley Longman Publishing Co., Inc.
- Are our jobs at risk? - 23.01.2024
- Are Our Jobs at Risk? - 10.01.2024
- Research Area Distributed Ledger Technology - 29.08.2023