by Konstantinos Pouliakas, Expert on Skills and Work, Cedefop
Artificial intelligence (AI) technologies are already having a profound impact on European labor markets, reshaping jobs and the professional skills demanded. Earlier waves of digital, AI-driven technologies (e.g. industrial autonomous robots) were more likely to affect routine jobs, mostly demanding people with medium-level skills or pay, supporting claims of technological automation exacerbating job polarisation. The recent advancement of generative AI technologies has further expanded the frontier of possible jobs that could be partially automated, with much research highlighting their immense capacity to augment human capabilities. Marked shortages for people in possession of advanced AI skills, for instance in machine learning, data science and AI ethics, are also evident in EU labour markets, hindering economic and productivity growth.
To investigate the extent to which AI technologies have proliferated within EU workplaces and how well-prepared European workers are for the AI revolution, Cedefop conducted a first representative European AI skills survey in 2024. Random samples of adult workers were drawn from eleven EU Member States, representing not only different geographical clusters but also countries with diverse digital exposures. Overall, information from more than 5 300 EU adult workers was collected. Based on this survey, Cedefop’s recent policy brief Skills empower workers in the AI revolution, which is part of its AI futures of work project, sheds light on how extensively AI is used, the impact of automation on job tasks, and the new training needs now facing much of the European workforce.
AI in the workplace: The Great AI divide
One of the most striking results of the Cedefop survey is that 28% of respondents use — or see coworkers use — AI tools in their workplace (Figure 1). More than 6 in 10 of those use AI to do their own job. AI is not being used yet in the workplaces of 55% of employees, suggesting a broad potential for further growth in AI adoption. While 17% are also not entirely certain if AI technology is used, which is indicative of the need for greater awareness and transparency when such tools are introduced in workplaces.
The new data reveals that there is a growing divergence in the frequency of use of AI technologies between the leading and lagging European countries. While over one third of workers in Northern European countries (e.g. Luxembourg, Belgium, Germany, France) say that they or their colleagues use AI technologies at their workplace, this is true for less than one in four workers in Poland (24%), Greece (21%) and Spain (16%), respectively. Meanwhile, Greece, along with Spain or Poland, have the lowest increase in addition to the overall lowest share of AI use in the workplace, showing a need for greater convergence in AI investments with other EU countries.
Part of such country differences are underpinned by differences in the industrial distribution of their economies, but differences in organisational or management practices and in worker skills and attitudes also come into play. AI use is quite widespread in the ICT (55%), professional (46%) and financial sectors (45%), but it is still low in health care (16%), transport (13%), agriculture (12%) and the accommodation and food services sector (11%).
Automation and unemployment
While many express fears about the potential adverse consequences of AI for employment and job quality, recent studies have tended to alleviate fears of massive job destruction.
A key finding of Cedefop’s survey is that 20% of European workers believe AI could take over more than half of their activities (Figure 2). Overall, the workers in the sample think that an AI system is already better equipped than a human to perform about one out of every four tasks of their job. Despite such marked scope for machine substitution, only 15% of workers in the EU11 sample are afraid that they may completely lose their job due to AI technologies. In some EU countries though, particularly Greece, about 1 in 4 workers express such concern. The survey validates previous analyses that have showed that Southern and Central and Eastern European countries are more vulnerable to the threat of automation.
Not all automation leads to job losses, however; it often involves reallocating tasks. Indeed, while 30% of people who use AI report that the technology has replaced some tasks they used to perform, 41% say they have incorporated new or different tasks into their role. A striking 67% indicate that AI has enabled them to do some tasks faster than before, highlighting the marked potential for efficiency and productivity gains that could be unleashed by AI.
Managing workers by algorithms
In addition to reshaping the task structure of work and unlocking efficiency dividends, there are concerns about workers’ algorithmic management — the use of machines and algorithms to allocate, monitor and reward job tasks — resulting in worker distrust and lower motivation to upskill and exert extra effort. Such algorithmic work is believed to have spread across workplaces following the growth in working-from-home arrangements during the Covid-19 pandemic. It is seen to be part of a general trend towards the wider ‘platformisation of work’ across diverse working environments, not only those in which job transactions take place in online labour platforms.
The new Cedefop data confirm that there has been a spread of “algorithmic” systems that automate work functions in European workplaces — from detecting faulty products to managing customer recommendations — with about one in seven adult workers usually working with digital tools or applications than can automatically do some of their tasks, using algorithms. For instance, 22% of workers are always or often using such automatic digital tools to recognise, translate, transcribe or generate text. And 20% use them to provide recommendations requiring specialised knowledge about their job.
Cedefop’s survey further corroborates earlier EC findings regarding the use by about 8% of European workers of other connected digital devices, such as smartwatches, trackers or sensors, as part of their job. Such devices are increasingly being used by employers for the purposes of monitoring workers’ attendance, vitals and job performance. Indeed, 16–17% already work with digital tools, applications or systems that either monitor or evaluate their job performance or provide task directions or instructions.
A sizeable 46% of European adult workers anticipate that AI will result in them being monitored more at their job. This is a marker of the potentially dire impact that AI may have on worker autonomy and well-being in future years.
New capabilities and training gaps
It is undeniable that the integration of new AI technologies in workplaces will demand new skills. 61% of respondents agree that they will need to acquire additional skills to handle AI’s impact on their job in the medium-term. 41% admit to having an AI skill gap, namely they need to further develop their knowledge and skills in using AI tools and systems for their job. What is concerning is that a notable 44% believe their employer will not provide such necessary training. This mismatch underscores the importance of bolstering learning pathways and upskilling strategies at both public and private, organisational level, so that all workers — regardless of occupation — can benefit from the digital transition. It demonstrates that need for closer employer-employee consultation and cooperation to ensure that AI is adopted within organisations in a transparent, fair, secure and human-centric manner.
Inclusion and equity challenges in VET
Cedefop’s AI survey further reveals concerns about AI widening gaps among different groups of workers. Middle-skilled, routine jobs are especially vulnerable to displacement by machines. The challenge, then, for policy is to assist the transition of people whose jobs are under threat with adequate upskilling, reskilling and social insurance schemes.
Despite significant AI skill gaps among workers, there is also great divergence in opportunities to train in AI-related knowledge, with only 15% of workers in the eleven EU countries participating in some education or training activity to further develop their AI skills in the last year. Such AI skills training remains less accessible for older and female employees and those under precarious employment contracts and in smaller-sized workplaces. All of this highlights the urgent need for inclusive policies that support workers at greater risk of being left behind. But it is also apparent that significant efforts need to be made to further upscale and customise the provision of AI skills training according to the needs of different sectors and occupations.
The productivity paradox and the need to reach AI’s blind spots
Despite high expectations for AI, 55% of its users in Europe say they have not yet seen a marked improvement in their job performance. Experts often explain this discrepancy by noting that AI adoption is still in its infancy, with many organizations lacking well-tailored work processes or robust training. Recent studies have tended to highlight that while AI has mostly facilitated productivity gains among novice and lower-skilled individuals, it has been less influential among higher-skilled professionals. Evidence of long-run aggregate productivity gains has also been weak.
Cedefop’s evidence highlights that overcoming gender and age-segregation associated with the use of AI technologies in labour markets will be crucial to facilitating potential AI productivity gains, given that females and older workers are less likely to deploy AI as part of their work.
Starting from the basics: The need to boost AI literacy
Using a unique scale for assessing individual’s AI literacy, the study highlights a low level of AI literacy among many adult workers in Europe. Between 40% and 62% of respondents report insufficient understanding of essential issues such as how algorithms function, the role humans play in developing AI, and how to identify potential biases or ethical risks. It follows that fostering a broad-based “AI culture” is crucial going beyond purely technical skills, to include the ability to interpret AI’s outputs, understand its limitations, and see its ethical and societal implications.
Toward an inclusive transition
For policymakers and company leaders, Cedefop’s survey data point to the urgency of promoting social dialogue and a robust supply of continuing training initiatives. Working with unions and employers helps validate these measures and accelerate AI adoption while ensuring automation is not merely a cost-cutting tactic. Cedefop’s and other evidence suggests that organizations with stronger employee participation institutions and well-structured training pathways show more positive AI outcomes.
Looking ahead: a new Cedefop AI skills foresight study
In addition to the AI skills survey, Cedefop has recently initiated a new AI skills foresight study. The initiative focuses on exploring mid-term scenarios of how automation and AI adoption will shape the future of organisations in strategic sectors of the EU economy. This research focuses on three key economic sectors — automotive manufacturing, geriatric nursing, and creative industries — which are likely to undergo marked restructuring in skill needs and potential sector-wide transformations over coming years because of their exposure to new AI technologies. The findings, to be published in 2026–27, aim to help policymakers and stakeholders guide an AI-driven transition that maximizes benefits while minimizing disruption for Europe’s workforce.
First preliminary results for the automotive sector highlight significant challenges for effective AI adoption in firms, including AI skill gaps, global competition and cost factors — exacerbated by geopolitical shifts. Cedefop’s early findings suggest that these industry changes will require upskilling a significant portion of the automotive workforce, requiring collaboration in new equipment, methods and innovation between private enterprise and public VET institutions.
An unstoppable transformation full of opportunities
All in all, AI is a powerful catalyst that goes beyond simple automation as it has marked scope to redesign production models and services, deepen knowledge transfer, foster innovation and worker autonomy and connect communities. The real challenge is to manage the transition so that it delivers benefits for European competitiveness, in line with the recent Draghi report, while avoiding the exclusion of any segment of the workforce. Training, re- and upskilling will prove vital to unlocking the potential of these technologies and for ensuring equitable gains for both businesses and workers. Cedefop’s research confirms we are witnessing a shift of enormous gravity — but also one that can pave the way for more modern, efficient, and equitable labour markets, provided that people and their skills remain at the centre of economic development strategy.