A recent study led by Anil Menon of the University of California, Merced, and Baobao Zhang of Syracuse University indicates that Americans’ attitudes toward potential job losses from artificial intelligence remain largely unchanged, even when they are told such disruptions could happen soon.
The research involved a survey of 2,440 U.S. adults in March 2024. Participants were randomly assigned to read one of several vignettes predicting that AI capable of replacing human workers could arrive in 2026, 2030, or 2060. A control group received no timeline information. The vignettes referenced rapid advances in large language models and generative AI systems like Genesis, Claude, and ChatGPT, with experts forecasting impacts on professions ranging from software engineering to nursing.
According to the findings, respondents who read about shorter timelines for transformative AI reported only slightly increased anxiety about automation’s impact on jobs. Their expectations for when job losses would occur and their support for government responses—such as retraining programs or universal basic income—remained essentially unchanged.
“These results suggest that Americans’ beliefs about automation risks are stubborn,” Menon and Zhang stated. “Even when told that human-level AI could arrive within just a few years, people don’t dramatically revise their expectations or demand new policies.”
The researchers noted that only those given the most distant timeline (2060) showed a significant increase in worry about job loss. They suggested this may be because respondents found a later date more credible than predictions of imminent disruption.
The study draws on construal level theory, which examines how perceptions of time influence risk judgments. Despite exposure to forecasts suggesting rapid change, participants did not show significant shifts in economic outlook or policy preferences related to automation.
Menon and Zhang argue that these results challenge the idea that making technological threats seem more immediate will lead to greater public support for regulatory measures or social safety nets. They also acknowledged limitations in their research design: it focused solely on timeline cues and did not examine other factors like beliefs about economic trade-offs or the credibility of expert opinions. Additionally, the survey does not track changes in individual perceptions over time.
“The public’s expectations about automation appear remarkably stable,” the authors said. “Understanding why they are so resistant to change is crucial for anticipating how societies will navigate the labor disruptions of the AI era.”
