Journalistischer Abstract:
Diese Hausarbeit untersucht die transformative Kraft der Künstlichen Intelligenz (KI) und ihre potenzielle Bedrohung für die europäische Integration. Es verknüpft die ökonomische Literatur zu den Auswirkungen von KI auf den Arbeitsmarkt mit der politischen Literatur zu den Effekten von Automatisierung auf Polarisierung und europäische Integration. Zunächst wird dargestellt, wie KI sowohl Routine- als auch komplexe Aufgaben automatisieren kann. Danach werden die historischen Auswirkungen von Automatisierung analysiert und Parallelen zur aktuellen Entwicklung der KI gezogen. Abschließend wird diskutiert, wie KI-induzierte wirtschaftliche Unsicherheit antieuropäische Stimmungen und Populismus verstärken könnte. Die Arbeit zeigt, dass gezielte politische Maßnahmen unerlässlich sind, um die durch KI verursachten Umwälzungen abzufedern – nicht nur im Interesse der Arbeitnehmer, sondern auch zur Sicherung der Stabilität der EU.
Abstract
This essay examines how artificial intelligence (AI) may affect European integration through its impacts on labor markets. Past waves of automation have fueled concerns about technological unemployment, yet the aggregate employment effects have usually proven more muted than predicted. AI, however, presents new capacities to automate non-routine cognitive and manual tasks. The essay argues AI may thus transform labor markets more profoundly than prior technologies. It traces how throughout history, periods of rapid economic transition and labor market disruption have triggered social unrest, increased populism and erosion of political centrism. AI’s disruption of work may similarly spur anti-EU sentiments by producing economic insecurity and inequality both within and between member states. Workers displaced by AI may resent European institutions perceived as both responsible for economic instability but unable to provide sufficient social protection. While AI might enhance long-run productivity, its impacts on labor may hamper integration in the near-term. Mitigating policies to support vulnerable workers will prove critical amidst AI-driven disruption, if stopping the current democratic backsliding and European disintegration is to be achieved. Though predictions remain assumption-heavy, the essay synthesizes multiple strings of research suggesting AI’s transformative potential could profoundly intersect with existing social and political tensions threatening the EU integration.
Introduction
1.000.000 users after five days.
100.000.000 monthly users today.
1.000.000.000 monthly visits.
These are the figures for ChatGPT, an AI language model that has far outstripped the growth of previous internet companies such as Netflix or Instagram (Duarte 2023). Within weeks of its release, AI seemed to go from sci-fi fantasy to coffee-table chatter. From rap battles between Shakespeare and Eminem, to solving complex scientific problems like the protein-folding problem (Service 2020: 1144), to posing potential existential risks (CAIS 2023): AI has characteristics that are at once comedic, utopian and dystopian, bringing it not only to the coffee table, but also to the world's central political arenas. While some view it as an overhyped wave of computerization, others see it as the ultimate innovation and the fourth industrial revolution. Understanding the extent of AI's potential and its difference to prior automations is crucial also in order to comprehend its impact on the stability of political bodies, such as the European Union. The economic impact of AI on the labor market is particularly critical, given the empirically close links between economic insecurity, rising populism, and rising Euroscepticism (Guiso et al. 2017: 5; Zhao 2022). Therefore, this essay intends to answer the research question:
How could AI systems affect the stability of the EU through their impact on the labor market? To answer this, I will:
1. Analyze how automation has historically impacted labor markets.
2. Compare how AI, as a fundamentally different technology, may affect labor differently than past automation waves.
3. Consider how AI's impact on the labor market could affect the EU's political stability through the demand for populism.
By doing so, this essay aims to unpack AI's complex relationship with European integration.
Labor Market & AI Definitions
To fully understand how AI could transform the labor market, it is first necessary to grasp what exactly artificial intelligence is, or can be. As described by Webb (2019: 35), artificial intelligence is often considered a machine learning algorithm that is able to take a given data input, such as textual data, and find methods to achieve a given output, such as being able to predict the next token of words in a sentence (what Large Language Models do). The ‘intelligence’ of AI refers to its ability to independently develop these methods and sub-goals to achieve a desired output, rather than following rigidly programmed rules like previous software. While humans code software, e.g. using a set of ‘what-if’ statements, the autonomous decision-making process is central to what makes AI artificially intelligent. As is the case with varying degrees of intelligence observed in animals and other organisms, AI systems can be categorized on different levels of intelligence based on capabilities and impact. I will refer to the categories of Narrow AI, AGI and HLAI throughout the text1. To categorize labor-force tasks, definitions from Autor, Katz and Kearney (2006) will be used: High skilled as non-routine cognitive activities, Medium skilled for routine cognitive/manual work, and Low skilled as non-routine manual tasks.
Historical automatization debate and actual effects
The possibility of automation displacing jobs has provoked debate for centuries. As early as the 19th century, Karl Marx raised concerns about industrial automation's impact on employment, theorizing it would create a reserve army of unemployed workers (Marx 1867). In fact, the evolution of technology has continually reshaped European labor markets over successive industrial revolutions. The first saw steam power mechanize production, displacing artisanal workers while enabling the transition to urban manufacturing and wage-labor. Later, electrification, mass production and assembly lines allowed consumer products to become affordable to the middle class, creating new roles in factories and cities. In the 20th and 21st century, debate renewed on automation's labor impacts with the computer revolution. Many prominent economists have frequently predicted technological anxiety and especially unemployment to be inevitable in the future (Mokyr et al. 2015: 31-32). In reality, many of these predictions of widespread unemployment didn’t manifest. Instead, the aggregate employment impact of automation has been more muted than predicted (Graetz and Michaels 2018: 32). Autor (2014: 41) seconds this and writes: “The onset of the weak U.S. labor market of the 2000s coincided with a sharp deceleration in computer investment—a fact that appears first-‐-order inconsistent with the onset of a new era of capital-‐-labor substitution”. One argument that did manifest, is that computerization predominantly substitutes for routine manual and cognitive tasks following explicit rules, but at the same complements and raises demand for non-routine manual or cognitive (e.g. problem-solving and complex communication) skills (Autor, Levy and Murnane 2003: 1322). Where demand is unsaturated, productivity enhancements can increase output of existing goods and services and by doing so increase employment. This is empirically backed: Analyzing 19 OECD countries from 1993-2007, Graetz and Michaels (2018: 6) found no relationship between increased industrial robot usage and overall employment rates, except for low-skilled workers. In the US, unemployment reached a 50-year low before COVID-19 despite surging tech capacity (Powell 2021). Looking deeper, Goos et al. (2009: 61) find European countries like Germany, Spain and Sweden experienced job polarization from 1993-2010, with growth in high-skill nonroutine cognitive occupations (e.g. managers, professionals) and low-skill manual occupations (e.g. cleaners, servers) but declines in routine middle-skill roles:
Confirming this, Jerbashian (2019: 18-19) found from 1993-2007 that lower robot and IT prices increased high-wage occupations but depressed routine middle-wage roles concentrated in automatable sectors, with no systematic evidence of impact on lowest paid occupations. Autor and Salomons (2018: 25-26) show that “although rising productivity reduces relative employment in the sectors in which it occurs, it augments employment in (downstream) customer sectors (as captured by the supplier effect) and boosts aggregate demand through its contribution to overall value-added. As with employment, the net effect on hours is strongly positive”.
In summary, automation and computerization have mostly exhibited a labor-polarizing effect, benefitting lots of high-skill abstract and low-skill manual jobs while substituting for mid-level routine tasks. The aggregate employment impact has proven positive and therefore much more modest and varied than early predictions. This history provides useful context for assessing the potentially transformative impacts of AI and machine learning technologies going forward.
How does AI automatization differ from previous automatization?
Automatization of Routine Tasks & Rising productivity
Like previous automation, Narrow AI programs are already well suited to replicate routine cognitive and manual tasks that follow codifiable procedures, continuing previous automatization trends for routine tasks. Another similarity is that AI in purely economic terms is expected to continue raising productivity: Goldman Sachs (2023) estimates generative AI alone could raise global GDP by 7%. However, as I will argue, AI goes beyond these impacts and through these differs from prior automation in fundamental ways suggesting more transformative impacts on the labor market and thus the political arena.
Overcoming Polanyi’s Paradox: Automatization of Non-Routine Tasks
First, AI is starting to overcome Polanyi's paradox - the inability of machines to perform tasks we consider creative and broadly intelligent. Polanyi (1966: 4) observed that: “We can know more than we can tell”, meaning that our practical knowledge of how the world works and how we interact in it exceeds our theoretical ability to define how it works. For computerization, this has meant that many non-routine tasks are non-codifiable, partly because we simply do not know why and how we do certain things and therefore cannot translate them into code. In the real world, this helps to explain why non-routine jobs, both at the low and especially at the high end of the skill spectrum, have been protected from automation. Now, with the expected emergence of increasingly capable AI systems, such as HLAI, overcoming this Paradox is increasingly likely and with it overcoming a barrier for replacing non-routine jobs (Bostrom 2017). The reason: AI doesn’t depend on human input in order to mirror human output.
As Brynjolfsson and Mitchell (2017: 1533) put it: “a much broader set of tasks will be automated or augmented by machines over the coming years. This includes tasks for which humans are unable to articulate a strategy but where statistics in data reveal regularities that entail a strategy”.
This allows AI to begin replicating non-routine tasks with the main limitations being computational power and robotic capability. With enough data, many more tasks are predictable than originally thought (Ford 2016: 111). One can already see the first applications, in which AI substitutes non-routine manual tasks as well as non-routine cognitive tasks. For example, AI can now simulate photography (Hsu and Myers 2023), generate paintings in the style of famous artists (Roose 2022), compose classical music (Schumann 2023), and write stories and poems that humans rate as creative. In the medical field, deep learning has achieved expert-level performance at breast cancer detection in multiple studies (Ehteshami Bejnordi et al. 2017: 2208; McKinney et al. 2020: 92-93) and diagnosing eye diseases from retinal scans (De Fauw et al. 2018: 1348-1349). This clearly demonstrates AI's emerging capability to replicate non-routine cognitive tasks and is consistent with Webb’s predictions in his quantitative study:
[…]| While individuals with low levels of education are somewhat exposed to AI, it is those with college degrees, including Master’s degrees, who are most exposed. Moreover, as might be expected from the fact that AI-exposed jobs are predominantly those involving high levels of education and accumulated experience, it is older workers who are most exposed to AI, with younger workers much less so (Webb 2019: 4).
He further finds that many low-skill manual occupations face high exposure to robots empowered by AI, although not as strongly as high skilled jobs. These findings don’t necessarily have to lead to job displacement. In sectors where demand is not yet saturated and price elasticity is high, we can expect AI to significantly reduce costs and change the way the product is delivered, but increase overall demand, so that overall employment could even increase in the short term until demand is saturated in the long term (Brynjolfsson and Mitchell 2017: 1533). For example, in healthcare, if AI takes over specific tasks previously performed by a doctor, the doctor's workload is reduced. Consequently, the doctor can see more patients and provide care at a lower cost per patient. However, this expansion of care may only be temporary, until demand is saturated.
Speed of creative destruction and labor replacement cycle
Apart from the scope of impact, a second key area in which AI differs from previous automatization is the speed of disruption. Disruption, however, does not necessarily imply negative consequences. Many technological disruptions have led to creative destruction and with it economic growth: Standard labor is destroyed by a more productive way of producing, opening up new areas in which labor has competitive advantages and thereby increasing overall productivity. With AGI or HLAI, the typical cycle of labor having a comparative advantage could be much shorter. Rapid advances in machine learning, driven by exponential growth in machine learning techniques, datasets and computing power, are reducing the time lag for artificial intelligence to require new complex capabilities, and thus lowering the barriers for AI to acquire advanced capabilities across domains through data training. Capability growth in AI usually follows power-law scaling relations, with a doubling of data leading to quadrupling of model ability (Bahri et al. 2021: 1). Therefore one may not just expect AI to break through Polanyi's paradox in a disruptive fashion, but do so at a faster speed. Even if we expect new jobs to be created, as has historically been the case (World Economic Forum 2020; The Economist 2023), we can also expect them to be destroyed again faster than in previous cycles of automation.
Accuracy of Labor Market Predictions
Thirdly, current predictions of significant labor market disruption from AI may prove more accurate than previous mass unemployment overstatements. Frey and Osborne (2017) estimated 47% of US jobs are at high risk due to automatability. Their study identified bottlenecks such as creative (e.g. originality and fine arts) and social (e.g. persuasion, deception, negotiation) intelligence, which could slow down unemployment effects. Since then, examples show creativity and abilities like persuasion moving from bottlenecks to capabilities for AI. Only recently, an AI system mastered the highly strategic game “Diplomacy”, which relies on actively deceiving opponents (Bakhtin et al. 2022: 7). To summarize this section, we can expect AI to continue to affect middle-skill, routine jobs that have already been affected by previous cycles of automation. While complementation has had, and substitution can have, productive long-term effects on an economy through the effect of creative destruction, AI is different in several key ways. Its impact is unlikely to be limited to middle-skill jobs, but rather, given the expected development of AGI and HLAI, may affect both high-skill cognitive tasks and low-skill manual tasks, potentially even more significantly. Moreover, we can expect the work cycle for creating new competitive advantages to be shorter. Finally, I argue that we can expect current predictions of automation and job displacement to be less exaggerated than earlier debates about automation because AI appears to have already solved several bottlenecks. Based on this comprehensive analysis of AI and its potential impact on the labor market, I will now turn to the implications for EU integration.
Impact on the EU
Current EU Integration
The EU already faces growing democratic backsliding and erosion of liberal norms in member states like Hungary and Poland (Roger Daniel Kelemen 2022: 178). As data from V-Dem shows, there has been a constant decrease in the share of liberal democracies within the last 15 years:
Meanwhile, the Eurozone crisis and austerity policies have fueled distrust of EU institutions, skyrocketing support for populist parties, and landmark events such as Brexit. This dynamic can be seen in the rise of anti-EU populist parties since the Eurozone crisis, with the two main right-wing parties, ID and EDR, holding around 18% of seats in the EU parliament and currently expected to increase in 2024 (Camut 2023). Moreover, the divergence of positions on European integration among member states has been accentuated (Kriesi 2016: 44).
AI’s general economic impact on EU integration
Economic history provides evidence that periods of labor market disruption can spur increases in anti-systemic politics and populism threatening social and political integration. Labor market shocks often undermine basic material security and stability for affected groups that trigger an identity-protective reaction among affected groups - attitudes that are usually negatively connected to EU support (Danilo Di Mauro and Memoli 2016). As Algan et al. (2017: 310) write: “There is a statistically and economically significant relationship between regional unemployment and a decline in trust toward the European Parliament and national parliaments”. Further, analyzing Western democracies between 1993 and 2016 Anelli, Colantone and Stanig (2019: 3) find that “Higher robot exposure at the individual level leads to poorer perceived economic conditions and well-being, lower satisfaction with the government and democracy, and a reduction in perceived political self-efficacy”. Additionally, Essletzbichler, Disslbacher and Moser (2018: 73) show in their three case studies that those regions “whose labour markets were exposed more and recovered less from the Great Recession, and those with high unemployment rates and benefit losses exhibited higher populist vote shares”. Identity threats by job insecurity are and have repeatedly been channeled into voting against establishment parties and policies.
Relatedly, theories link loss of socioeconomic status to demands for authoritarian populist leaders who scapegoat outsiders. A study found individual support for strong leadership is inversely related to personal income and country-level factors like inequality, wealth, and institutional quality (Chong and Gradstein 2018: 15). Those with less socioeconomic status reportedly desired tougher leadership more. However, support for populism is not purely due to economic reasons, but also stems from demands for identity and community in times of insecurity (Oesch 2008: 370). Authoritarian populists tap into a wider demand for cultural protectionism and national identity, especially during economic crises. Evidence shows rising far-right vote shares in regions most impacted by trade shocks and import competition, as Autor et al (2020: 3176-3177) found occurred in the US.
If the speed and scope of AI-driven labor market disruptions are faster and broader compared to previous automation waves as I have assumed and argued, the above analyses suggest this could strengthen authoritarian populist movements by threatening traditional socio-economic identities and mobilizing fears over status and identity loss. Rising populism in turn poses risks to consensual European integration as it often runs counter to it (Zhao 2022). Now we will take a closer look at the impact by job level.
AI’s job-related impacts
Mid-Skill Labor
I have first made the argument that AI will continue automatization of mid skill routine jobs. Two factors are key for applying this to EU integration: First, the substitutive effects within a country. Second, the differing degrees of exposition between EU countries. Both can be driving forces between anti-EU sentiments. As the OECD highlights, mid-skill jobs represent around 35% of employment in Greece, Slovenia and Italy, but only around 20-25% in Norway, Sweden, Netherlands and Denmark (OECD 2020: 225). As AI affects exposed sectors, this risks widening income gaps between higher-skill Western and lower-skill, lower resilience, high substitution risk economies. Furthermore, countries with less high-skill employment often have less resources to protect citizens through state involvement and redistribution, which is seen as central in order to dampen the impact of a potential labor market crisis, especially when trickle-down economic effects don’t materialize (Korinek and Stiglitz 2017: 3).
Low-Skilled Labor
As for low-skill occupations we may expect a lot more to be affected than traditionally assumed, due to the increasing ability of AI systems to replicate complex human perception and dexterity. If these predictions hold true, we can expect populist political dynamics similar to what we've seen in the past to continue, as AI-driven automation economically marginalizes portions of low-skilled labor. Populism often draws strength from economic marginalization of less-educated workers (Guiso et al. 2017: 22-23). The low-skilled facing job losses from automation lack skills for new complementary roles, making them susceptible to populist scapegoating narratives. With AI, they may come to view the EU as being responsible for economic insecurity but unable to provide sufficient social protection. Sensing opportunity, nationalist political parties can strategically scapegoat Brussels for disruptions caused by technology and globalization, while promising expanded welfare policies conditional on curtailing European integration, a common strategy already seen in Poland or Hungary, for example (Cook and Inglot 2021). Extensive automation impacts on low-skilled workers could potentially expand receptiveness to nationalist, Eurosceptic messaging and policy agendas proposing alternative solutions.
High-skilled labor
The literature is split on AI's impact on high-skilled labor - whether it will substitute or complement these jobs. In the short-run, I've argued AI will be more complementary as long as demand exists in high-skilled fields like medicine. In the long-run, substitution effects may strengthen. However, high-skilled workers can more easily retrain and adapt, creating new tasks. Whether the impact is complementary or substitutive may not make a big difference and both negatively impact EU integration in the short-term. If high-skilled jobs face risks, similar political resentments could emerge as with other displaced jobs, before alternatives emerge. Complementary effects could fuel inequality perceptions from lower-income groups, as innovator surpluses and intelligence enhancements are expected to accrue to high income jobs, which would widen initial inequality, not counting potential redistributive effects (Korinek and Stiglitz, 2017). Further, if new human tasks arise mainly in high-skill domains, this is also be expected to have inequality increasing effects (Acemoglu 2018: 1492). Inequality within states and between states runs counter to efforts to keep EU integration and is a central predictor for the success of populist parties. Stoetzer, Giesecke and Klüver (2021: 11-13) provide futher evidence that income inequality is a relevant driver for the electoral success of populist parties all over Europe. In that case, AI-related labor market polarization and EU-wide political polarization may be closely intertwined.
Predicting AI's precise impacts is challenging because many hypothetical assumptions are involved. However, research clearly suggests most jobs will face some degree of exposure to AI's effects, with uncertainty around specific outcomes. This uncertainty alone, from workers feeling uneasy about their own futures, could motivate votes for populist platforms expressing similar discomfort. Historically, periods of uncertainty tend to strengthen populist alignments. However, a technology's ability to replace jobs need not instantly lead to replacements, depending on political actions. Implementation and adoption speeds tend to be slower in sectors with heavy state involvement like education and healthcare. Governments can artificially delay changes through incentives prioritizing employment levels, as has often been the case (The Economist 2023). While predicting specific disruptions remains difficult, policy choices mitigating uncertainty and supporting retraining could alleviate potential political stresses from technological transitions that literature indicates will impact much of the workforce in coming years.
Is there also a case for stronger European integration?
While AI's disruption of labor markets clearly poses risks to European integration, countervailing forces may also strengthen unity and should be briefly mentioned to paint a more complete picture. Larger member states could see rising strategic value in pooling sovereignty at the EU level to effectively regulate cross-border technology giants. Uncoordinated national regulation of technologies can result in a regulatory 'race to the bottom' as firms threaten relocation to escape stringent rules, as OpenAI has done (Perrigo 2023b). It may therefore be in EU member states' collective interest to cooperate at a continental scale in order to set robust shared standards without ceding competitive advantages. As pioneering EU-wide initiatives on AI standardization, ethics and risk management show, such collaboration is possible and underway (European Commission 2021). Another argument to be made in favor of EU integration is through the long term economic benefits that may arise from AI. Regardless of the job displacement mechanisms, studies project sizable long-run productivity gains from generative AI adoption (Goldman Sachs 2023). If prosperity boosts are sufficiently shared and distributed, productivity could strengthen economic integration, trust and growth across the EU. Aligned with this potential, the discussion about Universal Basic Income, potentially financed through increased productivity gains from AI, has regained traction, especially among Silicon Valley leaders such as Sam Altman (Shead 2021). In general, both of the above arguments, the sovereignty pooling for cooperative regulation argument and the long term economic benefits argument face limitations. Firstly, any political initiatives requiring transfers of sovereignty could meet democratic resistance. Survey data on public support for the EU, while generally rising on average, show wide variations between countries, with, for example, only 50% of Greek respondents viewing the EU favorably compared to 83% in Sweden. (Fagan and Gubbala 2022). This, in line with increasing populist support, might make finding common ground for effective collective action in AI regulation hard to achieve. Further, even if we expect long term economic benefits and new jobs from AI, this may not necessarily be reflected in EU support given the short term focus of voting, which is more likely to be influenced by the short-term labor displacements that have been analyzed extensively. As Dassonneville (2016: 1) writes: “The increasing erosion in the linkages between voters and parties in advanced democracies has been seen as an indication that the longer term structural forces shaping voting behaviour that were identified by Campbell and his colleagues are giving way to shorter term determinants.”
Conclusion
This essay has delved into the complex relationship between AI, labor markets, and the stability of the European Union, revealing the multifaceted challenges and opportunities. In analyzing the historical impact of automation on labor markets, I've recognized that past waves of automation, while disruptive, resulted in more nuanced employment effects than the widespread unemployment initially feared. Instead, we observed a polarization of labor markets, with increases in high-skill non-routine cognitive and low-skill manual occupations, and a decline in routine middle-skill roles. Yet, AI presents a fundamentally different case. Unlike previous technologies, AI possesses the capacity to automate not just routine tasks, but also an expanding range of nonroutine cognitive and manual tasks. This potential for widespread automation could result in more profound labor market transformations than those caused by prior technologies. The implications of these transformations for the European Union are significant. AI-induced labor disruptions could amplify existing social and political tensions within and between member states, potentially fueling anti-EU sentiments, economic insecurity, and inequality. Workers displaced by AI might come to perceive European institutions as responsible for economic instability, yet incapable of providing adequate social protection, thus jeopardizing the process of European integration. At the same time, if properly managed, AI could improve long-term productivity, contribute to economic growth, and accelerate scientific progress, among many other benefits. It is therefore crucial for the EU to implement mitigating policies to support vulnerable workers and manage the transition toward an increasingly AI-driven economy. In conclusion, the intersection of AI, labor markets, and European integration is complex and filled with uncertainty. The future of AI is closely intertwined with the future of the EU project, and this essay has presented a rather pessimistic outlook. However, the future of AI and its impact on the labor market remains largely speculative, requiring ongoing research and adaptive policy-making.
Anonym
References
Algan, Y., Guriev, S., PAPAIOANNOU, E. and PASSARI, E. (2017). The European Trust Crisis and the Rise of Populism. Brookings Papers on Economic Activity, [online] pp.309–382. Available at: https://www.jstor.org/stable/90019460.
Anelli, M., Colantone, I. and Stanig, P. (2019). We Were the Robots: Automation and Voting Behavior in Western Europe. SSRN Electronic Journal. doi:https://doi.org/10.2139/ssrn.3428830.
Autor, D. (2014). Polanyi’s Paradox and the Shape of Employment Growth. [online] National Bureau of Economic Research. Available at: https://www.nber.org/papers/w20485.
Autor, D., Dorn, D., Hanson, G. and Majlesi, K. (2020). Importing Political Polarization? The Electoral Consequences of Rising Trade Exposure. American Economic Review, [online] 110(10), pp.3139–3183. doi:https://doi.org/10.1257/aer.20170011.
Autor, D. and Salomons, A. (2018). Is Automation Labor-Displacing? Productivity Growth, Employment, and the Labor Share. [online] www.nber.org. Available at: https://www.nber.org/papers/w24871.
Autor, D.H., Katz, L.F. and Kearney, M.S. (2006). The Polarization of the U.S. Labor Market. American Economic Review, 96(2), pp.189–194. doi:https://doi.org/10.1257/000282806777212620.
Autor, D.H., Levy, F. and Murnane, R.J. (2003). The Skill Content of Recent Technological Change: An Empirical Exploration. The Quarterly Journal of Economics, [online] 118(4), pp.1279–1333. Available at: https://www.jstor.org/stable/25053940.
Bahri, Y., Dyer, E., Kaplan, J., Lee, J.-H. and Sharma, U. (2021). Explaining Neural Scaling Laws.
Bakhtin, A., Brown, N., Dinan, E., Farina, G., Flaherty, C., Fried, D., Goff, A., Gray, J., Hu, H., Jacob, A.P., Komeili, M., Konath, K., Kwon, M., Lerer, A., Lewis, M., Miller, A.H., Mitts, S., Renduchintala, A., Roller, S. and Rowe, D. (2022). Human-level play in the game of Diplomacy by combining language models with strategic reasoning. Science. doi:https://doi.org/10.1126/science.ade9097.
Baum, S.D., Goertzel, B. and Goertzel, T.G. (2011). How long until human-level AI? Results from an expert assessment. Technological Forecasting and Social Change, 78(1), pp.185–195. doi:https://doi.org/10.1016/j.techfore.2010.09.006.
Bostrom, N. (2017). Superintelligence. [online] Google Books. Dunod. Available at: https://books.google.de/books?hl=de&lr=&id=eXg4DwAAQBAJ&oi=fnd&pg=PT249&dq=Superintelligence&ots= eo2JrIVmFj&sig=jydLcm0h9v5l61lDWPMRmBxGQhA#v=onepage&q=Superintelligence&f=false [Accessed 30 Aug. 2023].
Brynjolfsson, E. and Mitchell, T. (2017). What can machine learning do? Workforce implications. [online] Available at: https://www.cs.cmu.edu/~tom/pubs/Science_WorkforceDec2017.pdf. CAIS (2023). Statement on AI Risk |
CAIS. [online] www.safe.ai. Available at: https://www.safe.ai/statement-on-airisk.
Camut, N. (2023). Right wing set for big gains in 2024 EU election, polling shows. [online] POLITICO. Available at: https://www.politico.eu/article/european-election-2024-polls-right-wing-big-gains/ [Accessed 30 Aug. 2023].
Chong, A. and Gradstein, M. (2018). On the Preferences for Strong Leadership*. Social Science Quarterly, 99(4), pp.1267–1282. doi:https://doi.org/10.1111/ssqu.12489.
Cook, L.J. and Inglot, T. (2021). Central and Eastern European Countries. pp.880–898. doi:https://doi.org/10.1093/oxfordhb/9780198828389.013.50.
Danilo Di Mauro and Memoli, V. (2016). The Impact of ‘External’ Shocks on Attitudes Toward Europe: The Current Economic Crisis. doi:https://doi.org/10.1007/978-3-319-42954-0_3.
Dassonneville, R. (2016). Volatile voters, short-term choices? An analysis of the vote choice determinants of stable and volatile voters in Great Britain. Journal of Elections, Public Opinion and Parties, 26(3), pp.273–292. doi:https://doi.org/10.1080/17457289.2016.1158181.
De Fauw, J., Ledsam, J.R., Romera-Paredes, B., Nikolov, S., Tomasev, N., Blackwell, S., Askham, H., Glorot, X., O’Donoghue, B., Visentin, D., van den Driessche, G., Lakshminarayanan, B., Meyer, C., Mackinder, F., Bouton, S., Ayoub, K., Chopra, R., King, D., Karthikesalingam, A. and Hughes, C.O. (2018). Clinically applicable deep learning for diagnosis and referral in retinal disease. Nature Medicine, [online] 24(9), pp.1342–1350. doi:https://doi.org/10.1038/s41591-018-0107-6.
Duarte, F. (2023). Number of ChatGPT Users (2023). [online] Exploding Topics. Available at: https://explodingtopics.com/blog/chatgpt-users.
Ehteshami Bejnordi, B., Veta, M., Johannes van Diest, P., van Ginneken, B., Karssemeijer, N., Litjens, G., van der Laak, J.A.W.M., Hermsen, M., Manson, Q.F., Balkenhol, M., Geessink, O., Stathonikos, N., van Dijk, M.C., Bult, P., Beca, F., Beck, A.H., Wang, D., Khosla, A., Gargeya, R. and Irshad, H. (2017). Diagnostic Assessment of Deep Learning Algorithms for Detection of Lymph Node Metastases in Women With Breast Cancer. JAMA, 318(22), p.2199. doi:https://doi.org/10.1001/jama.2017.14585.
Essletzbichler, J., Disslbacher, F. and Moser, M. (2018). The victims of neoliberal globalisation and the rise of the populist vote: a comparative analysis of three recent electoral decisions. Cambridge Journal of Regions, Economy and Society, 11(1), pp.73–94. doi:https://doi.org/10.1093/cjres/rsx025.
European Commission (2021). The Act. [online] The Artificial Intelligence Act. Available at: https://artificialintelligenceact.eu/the-act/.
Fagan, M. and Gubbala, S. (2022). Positive views of European Union reach new highs in many countries. [online] Pew Research Center. Available at: https://www.pewresearch.org/short-reads/2022/10/13/positive-views-ofeuropean-union-reach-new-highs-in-many-countries/.
Ford, M. (2016). Book Review: The Rise of the Robots: Technology and the Threat of Mass Unemployment. International Journal of Human Resource Development: Practice, Policy & Research, 1(1), pp.111–112. doi:https://doi.org/10.22324/ijhrdppr.1.111.
Frey, C.B. and Osborne, M.A. (2017). The Future of employment: How Susceptible Are Jobs to computerisation? Technological Forecasting and Social Change, 114(1), pp.254–280.
Goldman Sachs (2023). Generative AI Could Raise Global GDP by 7%. [online] Goldman Sachs. Available at: https://www.goldmansachs.com/intelligence/pages/generative-ai-could-raise-global-gdp-by-7-percent.html.
Goos, M., Manning, A. and Salomons, A. (2009). Job Polarization in Europe. The American Economic Review, [online] 99(2), pp.58–63. Available at: https://www.jstor.org/stable/25592375 [Accessed 25 Jul. 2021].
Graetz, G. and Michaels, G. (2018). Robots at Work. The Review of Economics and Statistics, [online] 100(5), pp.753–768. doi:https://doi.org/10.1162/rest_a_00754.
Guiso, L., Herrera, H., Morelli, M. and Sonno, T. (2017). DEMAND AND SUPPLY OF POPULISM. BANKPEDIA REVIEW, 7(1-2), p.7. doi:https://doi.org/10.14612/guiso_herrera_morelli_sonno_1-2_2017.
Hsu, T. and Myers, S.L. (2023). Can We No Longer Believe Anything We See? The New York Times. [online] 8 Apr. Available at: https://www.nytimes.com/2023/04/08/business/media/ai-generated-images.html.
Jerbashian, V. (2019). Automation and Job Polarization: On the Decline of Middling Occupations in Europe. Oxford Bulletin of Economics and Statistics. doi:https://doi.org/10.1111/obes.12298.
Korinek, A. and Stiglitz, J.E. (2017). Artificial Intelligence and Its Implications for Income Distribution and Unemployment. [online] National Bureau of Economic Research Working Paper Series. Available at: https://www.nber.org/papers/w24174.
Kriesi, H. (2016). The Politicization of European Integration. JCMS: Journal of Common Market Studies, 54, pp.32–47. doi:https://doi.org/10.1111/jcms.12406.
Leontief, W. (1953). Domestic Production and Foreign Trade; The American Capital Position Re-Examined. Proceedings of the American Philosophical Society, [online] 97(4), pp.332–349. Available at: http://www.jstor.org/stable/3149288.
Marx, K. (1867). Das Kapital. Verlag Von Otto Meisner.
McKinney, S.M., Sieniek, M., Godbole, V., Godwin, J., Antropova, N., Ashrafian, H., Back, T., Chesus, M., Corrado, G.C., Darzi, A., Etemadi, M., Garcia-Vicente, F., Gilbert, F.J., Halling-Brown, M., Hassabis, D., Jansen, S., Karthikesalingam, A., Kelly, C.J., King, D. and Ledsam, J.R. (2020). International evaluation of an AI system for breast cancer screening. Nature, 577(7788), pp.89–94. doi:https://doi.org/10.1038/s41586-019-1799-6.
Mokyr, J., Vickers, C. and Ziebarth, N.L. (2015). The History of Technological Anxiety and the Future of Economic Growth: Is This Time Different? Journal of Economic Perspectives, [online] 29(3), pp.31–50. doi:https://doi.org/10.1257/jep.29.3.31.
Oecd and Ocde (2020). OECD Employment Outlook 2020 : Worker Security and the COVID-19 Crisis. Oecd Publishing; Éditions Ocde -07-07.
Oesch, D. (2008). Explaining Workers’ Support for Right-Wing Populist Parties in Western Europe: Evidence from Austria, Belgium, France, Norway, and Switzerland. International Political Science Review, 29(3), pp.349–373.
Polanyi, M. (1966). The tacit dimension. Chicago, Ill: The University Of Chicago Press.
Powell, J. (2021). Speech by Chair Powell on getting back to a strong labor market. [online] Board of Governors of the Federal Reserve System. Available at: https://www.federalreserve.gov/newsevents/speech/powell20210210a.htm#:~:text=The%20overall%20unemployme 18 nt%20rate%20was [Accessed 30 Aug. 2023]
Rodrik, D. (2017). Populism and the Economics of Globalization. SSRN Electronic Journal, 1(1-2). doi:https://doi.org/10.2139/ssrn.2992819.
Roger Daniel Kelemen (2022). Appeasement, ad infinitum. Maastricht journal of European and comparative law, 29(2), pp.177–181. doi:https://doi.org/10.1177/1023263x221097648.
Roose, K. (2022). A.I.-Generated Art Is Already Transforming Creative Work. The New York Times. [online] 21 Oct. Available at: https://www.nytimes.com/2022/10/21/technology/ai-generated-art-jobs-dall-e-2.html.
Roser, M., Herre, B. and Ortiz-Ospina, E. (2013). Democracy. [online] Our World in Data. Available at: https://ourworldindata.org/democracy.
Schumann, G. (2023). What Happens When A.I. Enters the Concert Hall. The New York Times. [online] 10 Jun. Available at: https://www.nytimes.com/2023/06/10/arts/music/ai-classical-music.html.
Service, R.F. (2020). ‘The game has changed.’ AI triumphs at protein folding. Science, [online] 370(6521), pp.1144–1145. doi:https://doi.org/10.1126/science.370.6521.1144.
Shead, S. (2021). Silicon Valley leaders think A.I. will one day fund free cash handouts. But experts aren’t convinced. [online] CNBC. Available at: https://www.cnbc.com/2021/03/30/openai-ceo-sam-altman-says-ai-couldpay-for-ubi-experts-disagree.html [Accessed 6 Jun. 2022].
Small, D.A., Loewenstein, G. and Slovic, P. (2007). Sympathy and callousness: The impact of deliberative thought on donations to identifiable and statistical victims. Organizational Behavior and Human Decision Processes, [online] 102(2), pp.143–153. doi:https://doi.org/10.1016/j.obhdp.2006.01.005.
Stoetzer, L.F., Giesecke, J. and Klüver, H. (2021). How does income inequality affect the support for populist parties? Journal of European Public Policy, 30(1), pp.1–20. doi:https://doi.org/10.1080/13501763.2021.1981981.
The Economist (2023). Your job is (probably) safe from artificial intelligence. [online] The Economist. Available at: https://www.economist.com/finance-and-economics/2023/05/07/your-job-is-probably-safe-from-artificialintelligence.
Webb, M. (2019). The Impact of Artificial Intelligence on the Labor Market. SSRN Electronic Journal. [online] doi:https://doi.org/10.2139/ssrn.3482150.
World Economic Forum (2020). Jobs of Tomorrow: Mapping Opportunity in the New Economy. [online] World Economic Forum. Available at: https://www.weforum.org/reports/jobs-of-tomorrow-mapping-opportunity-in-thenew-economy.
Zhao, J. (2022). The Rise of Populism and Implications on European Integration. Highlights in Business, Economics and Management, 2, pp.227–234. doi:https://doi.org/10.54097/hbem.v2i.2367.
Комментарии