Generative AI Anxiety Stems from Perceived Job & Resource Loss, Not Just Tech Adoption

Generative AI Anxiety Is Less About Using AI Than About Losing One’s Place

Discussions of generative AI anxiety often begin with a familiar explanation: technology is changing too quickly for people to keep up. Choi and Baek’s study moves the question elsewhere. Rather than focusing primarily on how often people use AI, it asks what people believe AI may take away from them.

The study proposes a pathway running from perceived occupational and social resource loss, through anxiety about adapting to technological change, and finally toward alienation from AI-related conversations. Its central framework is therefore not merely one of technological difficulty, but of perceived loss: jobs, expertise, status, opportunity, and one’s position within a changing social order.

Anxiety Begins With Resources That May Disappear

One of the study’s strongest conceptual moves is its use of Conservation of Resources theory. A job is not only a source of income. It also carries accumulated expertise, professional identity, social status, and a sense of competence.

From this perspective, AI does not have to replace someone’s job for anxiety to emerge. The possibility of replacement can already signal a potential loss of resources. The authors therefore define AI anxiety not simply as stress caused by complicated technology, but as an emotional response to possible losses in occupational and social resources.

The regression results support this framing. Among 667 respondents, frequency of generative AI use did not significantly predict technological adaptation anxiety. By contrast, three variables did: perceived replacement of jobs across society (β=.230), perceived replacement of one’s own job (β=.186), and perceived widening of socioeconomic inequality (β=.104). Concerns about declining human creativity and loss of human control over AI were not significant once the other variables were considered together.

This shifts the map of AI anxiety. Questions such as whether AI will become more creative than humans or escape human control may dominate public discussion, yet in this dataset the concerns more closely tied to anxiety were much more immediate: Will jobs disappear? Could my own work be replaced? Will the benefits of AI become concentrated among those who already possess capital and technology?

Society and the Self Matter More Than the Jobs of People Nearby

The study separates perceived job replacement into three levels: replacement across society, replacement affecting family and friends, and replacement of one’s own work.

The distinction produces an interesting pattern. Worry that acquaintances or family members might lose their jobs was not a statistically significant predictor of anxiety after other variables were controlled. Perceptions that many jobs across society would disappear, however, remained significant. So did the possibility that one’s own job could be replaced.

These two levels sit at opposite ends of the scale: one is societal, the other personal. Yet both force the individual to recalculate their position. Headlines announce that whole categories of work may disappear, while at the desk an AI system completes tasks that previously required hours of human labor. The anxiety identified in this study sits between those two scenes.

Alienation Appears at the Conversation Table

One of the study’s most distinctive ideas is its definition of AI-related conversational alienation.

Rather than treating alienation only as an abstract philosophical condition, the researchers operationalize it through an everyday situation: people discussing generative AI while someone feels unable to follow the flow of the conversation. The survey asks whether respondents have experienced a sense of emotional alienation because they could not keep up with discussions about generative AI.

Here, anxiety becomes the key connecting variable.

The correlation between anxiety and conversational alienation was r=.501, one of the strongest relationships among the variables examined. When anxiety was added to the regression model predicting alienation, the model’s explanatory power rose from 19.1% to 32.3%. Anxiety itself showed a substantial positive association with alienation (β=.423).

The relationship between perceived societal job replacement and alienation also changed once anxiety entered the model. Before anxiety was included, perceived societal replacement significantly predicted alienation (β=.177). After anxiety was introduced, the coefficient fell to β=.079 and was no longer statistically significant. The effect of perceived replacement of one’s own job also declined, from β=.174 to β=.096.

The authors interpret these shifts as evidence consistent with an indirect pathway: perceptions of occupational replacement may contribute to adaptation anxiety, which in turn is associated with feeling excluded from AI-related conversations.

Frequent Users Were Not Less Anxious, but They Felt Less Left Out

Another result complicates the familiar assumption that simply using AI more often will solve AI anxiety.

Frequency of generative AI use did not significantly reduce adaptation anxiety. Familiarity with the tools was therefore not enough to remove the feeling that technological development was moving faster than oneself.

Yet greater AI use was associated with lower conversational alienation. In the correlation analysis, usage frequency was negatively associated with alienation (r=-.110), and the negative relationship remained in the regression model that included anxiety.

This suggests that the divide emerging around generative AI cannot be reduced to a simple distinction between users and non-users. Several layers operate simultaneously: whether someone can use AI, whether they believe their occupation remains viable, whether they possess the resources required to adapt, and whether they can participate in the social conversations surrounding technological change.

What the Final Tables Reveal

Reading Tables 4 and 6 together makes the study’s argument especially clear.

Table 4 shows that AI anxiety is more strongly associated with perceptions of occupational replacement and widening inequality than with frequency of AI use itself. Table 6 then shows that anxiety becomes a major predictor of AI-related conversational alienation, while the direct relationship between some replacement perceptions and alienation becomes weaker once anxiety is introduced.

The resulting picture of generative AI anxiety is therefore less about fear of operating a new machine and more about uncertainty over one’s position within a changing occupational and social landscape.

The study’s most compelling implication emerges from an ordinary scene: a group of people discussing what AI can now do, which tools they use, and how work is changing. Someone at the table has heard the terminology but cannot follow the exchange. Behind that silence may sit another calculation: whether their skills still matter, whether their job will remain, and whether the gap between those who can adapt and those who cannot is already widening.

The new divide created by generative AI may therefore appear not only on the computer screen, but also in who can remain inside the conversation.