A new representative survey from TU Darmstadt, conducted with YouGov among more than 2,000 people in Germany, reveals a striking paradox: the more people understand artificial intelligence, the more they expect it to replace them at work. Around 43 percent of respondents with very strong AI knowledge said they believe AI could soon take over their job, a significantly higher share than among those with less familiarity with the technology.
The findings are part of the second edition of the study, which measures changes in German attitudes toward AI over the past year. The researchers described the pattern as both remarkable and worrying, noting that people who truly grasp what AI can do are also the most aware of its disruptive potential.
Key facts from the study
- More than 2,000 people in Germany were surveyed in collaboration with YouGov.
- The survey is representative of the German adult population.
- 43% of respondents with very strong AI knowledge expect AI to take over their work.
- Only 15% of all respondents have received any AI training.
- Hallucinations and data protection top the list of concerns.
- Only 40% believe superintelligence is likely.
- Younger workers are more anxious than older workers.
- Concern about job security rose across every occupational group since last year.
Knowledge breeds fear, not confidence
Conventional wisdom might suggest that learning more about a technology would reduce fear of it. This survey turns that assumption on its head. The people who understand AI best are not reassured by their knowledge; instead, they see clearly how quickly the technology is advancing and how many tasks it can already perform. They are also more likely to recognize the potential for AI systems to improve, which makes them more realistic about the threat to their own livelihoods.
Peter Buxmann, who led the study, called the pattern both remarkable and worrying. He pointed out that even among experts, there is no consensus that AI will simply create more jobs than it destroys. Instead, those who know the most are bracing for disruption. This challenges the upbeat narrative often heard from tech companies and policymakers who argue that AI will free workers to focus on higher-value activities.
The researchers also found that industry matters less than output. If a job primarily involves producing text, analysis, reports, code, or slides, the worker is more exposed to automation. This suggests that white-collar roles, which were once considered safe, are now at the forefront of AI disruption.
Practical fears outweigh existential ones
Another surprise from the study is that the biggest worries are not about machines becoming superintelligent or taking over the world. Instead, respondents are most concerned about hallucinations and data protection. Hallucinations refer to instances where AI systems generate convincing but completely false information, which can be dangerous in professional settings. Data protection concerns are also widespread, especially in Germany, where privacy is a particularly sensitive issue.
Only 40 percent of respondents believe that superintelligence is even likely. This indicates that most people are more focused on the immediate risks of using AI tools than on distant science-fiction scenarios. The fact that practical concerns rank higher than existential ones suggests that the public is becoming more sophisticated in its understanding of AI, even while its knowledge remains unevenly distributed.
Younger workers are more anxious
The study also found that younger workers are more anxious about AI than older workers, which inverts the usual expectation that older employees would be more threatened by new technology. One possible explanation is that younger workers are more exposed to AI in their education and early careers, making them more aware of its capabilities. They are also more likely to be working in roles that involve the kind of output AI can generate, such as writing, coding, or data analysis.
Concern about job security rose across every occupational group since the previous year’s edition of the study. This broad increase suggests that AI anxiety is not limited to any one sector or demographic. It is becoming a mainstream workplace issue, and employers who ignore it are doing so at their own risk.
The training gap
One number should trouble employers more than any other: just 15 percent of respondents have had any AI training at all. This is particularly striking because Europe has had a rule about AI literacy for eighteen months. Article 4 of the European Union’s AI Act has been in force since February 2, 2025. It obliges providers and deployers of AI systems to take measures to ensure that staff who operate those systems have an adequate level of AI literacy, weighed against their training, technical knowledge, and the context in which the AI is used.
The regulation does not set a specific level of literacy and only applies to organizations that actually use AI systems. Therefore, the 15 percent figure is not a direct measure of compliance. Still, it is a useful readiness indicator. After eighteen months of a legal duty to promote AI literacy, most German workers have received no training at all. This suggests that the AI Act has not yet translated into meaningful workplace education.
The lack of training is especially concerning given the survey’s finding that knowledge is linked to realistic perceptions of risk. If workers do not understand how AI works, they may not be able to protect themselves from its risks or advocate for the right kind of support. They may also be more vulnerable to both overconfidence and panic.
What this means for employers and policymakers
The findings carry a clear message for employers: AI training is no longer optional. It is a necessary investment in workforce stability and mental health. Training should not be a one-time lecture or a generic online course. It should be tailored to the specific ways AI is used in each role and should include honest discussion of both the benefits and the risks. Workers need to know what AI can do, what it cannot do, and how their jobs may change.
Employers should also be prepared to give honest answers about the potential for job displacement. The study suggests that reassurance without substance will not work. Workers are already anxious, and they are likely to see through empty promises. What they want is realistic information and a clear plan for how they will be supported in a changing workplace.
Policymakers, meanwhile, need to consider whether the AI Act’s literacy provisions are strong enough. The fact that only 15 percent of German workers have received AI training suggests that soft obligations are not sufficient. Stronger incentives, clearer standards, or even enforceable minimum requirements may be needed to close the gap between legal ambition and reality.
There is also a broader question about the role of social safety nets. If AI does lead to significant job losses, training alone will not be enough. Workers will need income support, career counseling, and assistance with transitions to new roles. The study highlights that losing a job costs more than income. It costs identity, purpose, and social connection. That is a loss that no algorithm can replace.
The survey’s findings add to a growing body of evidence that AI is reshaping the labor market in ways that are more complex than the headline narrative suggests. In Europe, the impact of AI on jobs looks different from the tech-industry vision of seamless augmentation and new opportunities. The reality is that many workers are worried, and their worries are grounded in a clear-eyed assessment of what AI can do.
The clearest takeaway from the TU Darmstadt study is that ignorance is not bliss, but knowledge is not the same as security. The people who know AI best are the people who fear it most, and that should be a signal to everyone who cares about the future of work. Understanding AI is the first step, but it must be followed by action, training, and honest conversations about the future.
Until then, the gap between AI’s capabilities and workers’ readiness will only grow wider. And the anxiety that already exists will likely intensify, especially for younger workers who are just starting their careers and facing a labor market that is changing faster than institutions can adapt. The burden is now on employers, policymakers, and educators to turn awareness into preparedness, before the fear becomes self-fulfilling.