📄 social science

Less Afraid, Still Wary: Europeans’ Fear of AI-Driven Job Loss Before and After Generative AI

Analyzing Eurobarometer data from 2017 and 2024 across 27 EU countries, this study reveals that perceived AI-driven job threats have surprisingly declined despite the rise of generative AI, with acceptance driven more by perceived benefits than fear and influenced by factors such as digital skills, employer transparency, and training.

Original authors: Amit Vishwakarma, RAJENDER KUMAR GUNEJA

Published 2026-10-14

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Amit Vishwakarma, RAJENDER KUMAR GUNEJA

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

For decades, the story of artificial intelligence in the workplace was a simple one: machines would take over repetitive, manual tasks, leaving humans to handle the complex thinking. This narrative suggested that the more skilled a worker was, the safer they would be from automation. However, the arrival of a new kind of technology, known as generative artificial intelligence, has rewritten that script. Unlike earlier tools that followed strict rules, this new technology can write code, draft letters, and summarize reports, effectively stepping into the realm of white-collar work. This shift has sparked a global conversation about whether these tools are helpers or replacements. To understand how people are reacting, researchers look at two main ideas. First, they examine the feeling of job insecurity, which is the worry that a machine might take one's livelihood. Second, they study how people decide to accept or reject a new technology, a decision often driven by whether they see it as useful or dangerous. Understanding these feelings is crucial because fear can drive people to demand strict rules, while a sense of usefulness can lead to adoption, regardless of the risks.

A team of researchers set out to see how these feelings changed across Europe between 2017 and 2024, a period that saw the sudden rise of generative AI. They analyzed data from millions of people in twenty-seven European Union countries, comparing surveys taken before the new technology arrived with those taken after. Their goal was to determine if the fear of losing jobs to machines had grown, and to understand what factors, such as a worker's digital skills or their employer's communication style, influenced that fear. They also wanted to know if people who feared AI were more likely to want new workplace rules, and whether those who saw benefits in the technology were more willing to use it.

The results offered a surprising twist. Contrary to the expectation that the arrival of powerful new tools would make people more anxious, the fear of job loss actually decreased across almost every country and social group. In 2017, a large majority of respondents believed that robots and AI would steal more jobs than they would create. By 2024, that number had dropped significantly. This decline happened even though the new technology was capable of doing the very tasks that skilled office workers perform. The researchers found that this drop in fear was not simply because unemployment rates had fallen; the trend held true even when accounting for the health of the local economy. It suggests that as people became more familiar with the technology and saw it being used to help rather than replace, their immediate panic subsided.

However, the relationship between a worker's skills and their fear changed in a way that was not anticipated. In 2017, people with strong digital skills felt less threatened, likely because they felt confident in their ability to adapt. By 2024, this pattern had flipped. Among employees in 2024, those with higher digital skills actually reported feeling slightly more threatened, once their general views on technology were taken into account. The researchers suggest this might be because skilled workers understand the new technology better and, therefore, see more clearly exactly what it is capable of replacing. It is a case where knowing more about the tool makes one more aware of the potential danger, rather than less.

The study also highlighted how employers can influence these feelings, but not in the way one might expect. The researchers found that when employers were transparent about how they used AI to manage work, their employees felt less threatened. Knowing that the technology was being used, and having it explained, reduced the uncertainty that often fuels fear. However, this transparency did not necessarily make employees love the technology or view it positively. On the other hand, when employers provided training to help workers use these new tools, it did not reduce fear, but it did make employees more accepting of the technology in their daily work. This suggests that training helps people see the practical benefits, which is a stronger driver of acceptance than simply being told the technology is safe.

Ultimately, the study revealed that people can hold two conflicting views at once. An employee could believe that AI is dangerous to jobs while simultaneously believing it is useful for getting work done. The research showed that the decision to accept AI at work depended far more on whether the worker saw it as beneficial than on whether they were afraid of it. Even those who were worried about job loss were not necessarily against the technology; instead, they were the ones most strongly in favor of creating rules to protect privacy and ensure human oversight. This indicates that the demand for regulation is not just a reaction to fear, but a desire to manage a powerful tool that many people find valuable but want to keep in check. The findings suggest that to build a workforce that is both confident and productive, employers should focus on showing the concrete benefits of AI while being open about how it is used, rather than trying to simply reassure workers that their jobs are safe.

Technical Summary: "Less Afraid, Still Wary: Europeans' Fear of AI-Driven Job Loss Before and After Generative AI"

Problem Statement
While generative artificial intelligence (GenAI) has rapidly integrated into everyday work, it remains unclear whether this technological shift has intensified or alleviated workers' fears regarding job displacement. Previous research on automation attitudes (e.g., Gnambs and Appel, 2019) often predates GenAI, and recent studies lack comparative baselines or fail to treat perceived job threat as a central outcome. Furthermore, existing literature often relies on small convenience samples rather than representative data, and few studies have examined how employer practices (transparency and training) interact with digital skills to shape threat perceptions and technology acceptance. This study addresses three gaps: (1) determining if perceived job threat changed between 2017 and 2024 across the EU; (2) analyzing how digital skills and employer practices relate to threat; and (3) understanding how threat and perceived benefits drive AI acceptance and support for workplace regulations.

Methodology
The study employs a three-stage quantitative design using data from two representative Eurobarometer surveys covering 27 European Union countries:

Data Sources: Eurobarometer 87.1 (March 2017, $N=26,555$ ) and Eurobarometer 101.4/Special Eurobarometer 554 (April–May 2024, $N=26,404$ ). The UK was excluded from the 2017 dataset to ensure comparability of the 27 member states.
Measurement: Constructs included digital skills, evaluation of digital technology, perceived job threat, perceived benefits of AI, employer transparency, employer training, view of AI at work, and support for workplace rules. Items were measured on 4-point scales.
Analytical Approach:
1. Measurement Invariance: Ordinal Confirmatory Factor Analysis (CFA) using the Weighted Least Squares Mean- and Variance-adjusted (WLSMV) estimator tested configural, threshold, metric, and scalar invariance across years to ensure latent means were comparable.
2. Structural Equation Modeling (SEM): Multi-group SEM tested hypotheses across years and within the 2024 employee subsample. Fixed-effects regression models controlled for country-level variables (unemployment, inequality, GDP) and individual demographics.
3. Artificial Neural Networks (ANN): Multilayer perceptrons with one hidden layer were used to detect non-linear effects and rank predictor importance via permutation on test folds (10-fold cross-validation).
Theoretical Framework: The study integrates Appraisal Theory, Technology Threat Avoidance Theory (TTAT), and the Technology Acceptance Model (TAM), treating threat and benefit as distinct appraisals.

Key Results

Trend in Fear (RQ1): Perceived job threat from AI declined significantly between 2017 and 2024. Latent threat scores dropped by 0.28 standard deviations. This decline was observed in 25 of 27 countries and across all social groups (education, age, occupation). Fixed-effects regression indicated that falling unemployment explained only a small fraction (approx. one-eighth) of this decline, suggesting the change is not merely a reflection of improving labor markets.
Role of Digital Skills (H1): The relationship between digital skills and job threat reversed. In 2017, higher skills were associated with lower threat ( $\beta = -0.09$ ). By 2024, holding evaluations of technology constant, higher skills were associated with higher threat ( $\beta = 0.06$ in the pooled model; $\beta = 0.37$ in the employee model). This suggests a suppression effect where skilled workers, aware of AI's capabilities, perceive greater specific risks.
Employer Practices (H2, H3):
- Transparency: Employer transparency regarding AI use was strongly negatively associated with perceived job threat ( $\beta = -0.21$ ) but had no direct significant effect on the overall view of AI at work.
- Training: Employer training was positively associated with a favorable view of AI at work ( $\beta = 0.13$ ) but did not significantly reduce perceived job threat.
Drivers of Acceptance and Regulation (H4–H6):
- Acceptance of AI at work was driven far more by perceived benefits ( $\beta = 0.64$ ) than by the absence of threat ( $\beta = -0.22$ ).
- Perceived job threat was positively associated with support for workplace rules ( $\beta = 0.33$ ). Notably, support for rules was also high among those who perceived benefits, indicating that demand for regulation is not solely fear-driven.
ANN Findings: The ANN models confirmed that perceived benefits were the strongest predictor of a favorable view of AI, while transparency was the most important predictor for reducing threat. The models did not overfit, and non-linear effects did not significantly outperform the linear SEM.

Key Contributions

Longitudinal Comparison: This study addresses a gap in the literature by comparing representative EU samples before and after the arrival of GenAI using identical questions and formal tests of measurement equivalence, demonstrating that fear of job loss has actually decreased despite the technological shift.
Differentiation of Employer Practices: The study distinguishes between two employer practices, showing that transparency primarily mitigates fear (uncertainty reduction), while training primarily fosters acceptance (utility realization).
Methodological Rigor: The application of a combined SEM–ANN design to large, representative national samples, moving beyond the small convenience samples common in technology adoption research.
Reversal of Skill-Threat Link: The finding that digital skills, once a buffer against fear, are now associated with higher threat perceptions (when controlling for general tech evaluation) challenges previous assumptions about the protective nature of skills in the GenAI era.

Significance and Claims
The paper claims that the arrival of generative AI has not deepened European fears of job loss; rather, these fears have receded, likely due to familiarity and early evidence of productivity gains rather than mass displacement. The study argues that the role of digital skills has shifted: while they previously signaled adaptability, they now may signal a clearer understanding of AI's disruptive potential.

From a policy and management perspective, the authors claim that explaining AI use (transparency) is crucial for reducing fear, while demonstrating concrete benefits and providing training are essential for building acceptance. Crucially, the demand for workplace AI regulations appears robust across the spectrum of opinion, driven by both those who fear AI and those who welcome it. Therefore, regulatory frameworks (such as the EU AI Act) are unlikely to face public opposition even if job-loss fears subside. The study concludes that fear of AI-driven job loss behaves more like a shared social belief than a personal calculation, suggesting that public information regarding actual labor market effects may be as important as individual retraining programs.

Technical Summary: "Less Afraid, Still Wary: Europeans' Fear of AI-Driven Job Loss Before and After Generative AI"

More like this