AI knowledge may increase job fears rather than reduce them, research finds

People who know more about artificial intelligence may be more worried about losing their jobs to the technology, according to new research from TU Darmstadt. The findings challenge the common assumption that anxiety about AI mainly comes from a lack of understanding about how the technology works.

The study, published as part of the AI Monitor 2026, is based on a representative survey of more than 2,000 people in Germany. It found that employees with a stronger understanding of AI were more likely to believe that the technology could eventually take over parts of their work.

Around 43% of respondents with very strong AI knowledge said they expected AI could soon take over their jobs. The figure was considerably lower among people with less knowledge of the technology. The results suggest that familiarity with AI may make its potential impact on employment more visible rather than making workers feel more secure.

Professor Peter Buxmann, who led the study, described the finding as both remarkable and concerning. Discussions around AI-related fears often suggest that better information will make people less anxious, but the research points in the opposite direction.

The type of work people perform also appears to be important. Roles that mainly produce digital material, such as written content, reports, analysis, computer code or presentations, may be more exposed to AI than jobs that depend heavily on physical activity or a person’s presence in a particular location.

However, the research also indicates that AI adoption at work remains relatively limited. Despite the rapid development of generative AI tools and growing interest from businesses, many employees are not using AI regularly as part of their daily work.

AI skills remain limited among workers

The study found a significant gap between growing awareness of AI and practical training. Only 15% of respondents said they had taken part in some form of AI training, suggesting most people are being asked to adapt to the technology without much formal preparation.

That lack of training could make uncertainty more difficult to manage. Employees may understand that AI can handle an increasing range of digital tasks while having limited guidance on how the technology will affect their responsibilities, career prospects, or workplace expectations.

Respondents’ concerns were not limited to job losses. AI hallucinations, in which systems produce inaccurate or fabricated information, and data privacy were among the leading concerns. These issues reflect practical risks that workers and organisations are already encountering as AI tools become more common.

The research also found that relatively few people viewed the arrival of a future superintelligence as likely. Only about four in 10 respondents considered such a scenario probable. This suggests that concerns about AI are more closely tied to its current and near-term effects than to more extreme predictions about the distant future.

Job insecurity has also become more widespread. Compared with the previous year, concern about employment increased across all occupational groups included in the research. Younger workers reported particularly high levels of anxiety about their jobs, indicating that the impact of AI is being felt differently across age groups.

The findings point to a complicated relationship between AI knowledge and confidence. Learning more about the technology can help people understand its capabilities, but that understanding may also make the potential for workplace disruption harder to ignore.

Training could be more effective than reassurance

The researchers argue that simply telling workers not to worry about AI is unlikely to address the underlying problem. Job losses can have consequences that extend beyond a person’s financial situation, affecting social relationships, identity and a sense of purpose.

Buxmann said that people who lose their jobs lose more than income, as employment can also provide social connection and a sense of purpose. The finding suggests that discussions about AI and employment need to consider the wider consequences of workplace disruption rather than focusing only on productivity or economic gains.

Better training could therefore play a central role in helping workers adapt. Providing employees with practical knowledge about AI could give them a clearer understanding of what the technology can and cannot do, while also helping them identify ways to use AI within their existing roles.

At the same time, training alone may not eliminate fears about employment. If some tasks become automated, workers could still face changes to their responsibilities or reductions in demand for certain skills. Employers and policymakers may therefore need to combine AI education with broader plans for workforce development and career transitions.

The research also highlights a potential mismatch between the creation of new AI-related roles and the jobs that could disappear. Although generative AI is expected to create new types of employment, Buxmann argued that these new roles are unlikely to replace disappearing jobs on a one-to-one basis.

For businesses, the findings suggest that preparing employees for AI-driven change may be more useful than trying to reassure them that their jobs are safe. For policymakers, the results highlight the need for realistic discussions about which forms of work are most vulnerable and how to support affected workers.

Rather than treating AI anxiety as a problem caused by ignorance, the research presents it as a potentially informed response to a rapidly changing workplace. As employees better understand what AI can do, their concerns may increase because they can see more clearly how the technology could affect their livelihoods.