Recent research published in quarterly public opinion This provides evidence that the public has deeply divided beliefs about the economic impact of artificial intelligence. This study suggests that these differing opinions tend to align with existing political divisions in North America. This finding shows that politicians can easily exploit these divisions to turn technology policy into a highly polarized political issue.
The research team included Sophie Bowin, Beatriz Magistro, R. Michael Alvarez, Bart Bonikowski, and Peter J. Loewen. These authors represent universities in the United States and Canada, including the University of British Columbia, Northeastern University, California Institute of Technology, New York University, and Cornell University.
New technologies often bring overall economic benefits to entire countries. At the same time, the tangible costs and benefits of these advances are rarely evenly distributed among the population. This unequal distribution tends to make technological progress a polarizing political issue. Some groups may experience increases in wealth and employment opportunities, while others may face unemployment and decreases in wages.
The authors wanted to understand the general public’s causal beliefs regarding artificial intelligence. Causal beliefs refer to the ways in which people logically connect certain causes and certain effects in their minds. In this case, the researchers wanted to know how individuals thought new software would directly impact different parts of the economy. They sought to uncover who the public sees as the inevitable winners and losers of this emerging technology.
Understanding these beliefs is especially important because artificial intelligence represents a unique type of technological change. Past technological revolutions, such as the rise of factory machinery, primarily affected manual labor. In contrast, modern automated systems are capable of performing complex cognitive tasks. This means that office workers, writers, and managers may feel more financially vulnerable than ever before during technological change.
Knowing how people form these causal beliefs helps experts predict future political behavior. When individuals believe that a particular group is being unfairly harmed by a new policy or technology, they are more likely to support politicians who promise to stop that harm. The authors set out to determine whether these attitudes toward technology are already taking shape in voters’ minds. They predicted that these early beliefs would provide the basis for future political movements.
To investigate this topic, researchers designed a new survey instrument. A survey instrument is a structured questionnaire used to collect specific data from a large number of people. They distributed the survey to a sample of approximately 6,000 adults living in the United States and Canada. This large sample size allowed the authors to capture a wide range of perspectives across different regions and economic backgrounds.
The survey asked respondents a series of questions about how artificial intelligence will impact various aspects of society. Questions covered everyday topics such as consumer prices, product quality, and the overall job market. The researchers specifically asked participants whether they thought this technology would complement or completely replace workers. Complementing the worker means that technology acts as a tool to make that person’s job easier and more efficient.
Replacing workers means that technology does the job completely on its own, leaving human workers unemployed. After collecting thousands of survey responses, the researchers analyzed the data using a technique called latent class analysis. Latent class analysis is an advanced statistical technique that helps researchers discover hidden groups or classes within large complex datasets. It works by observing how people answer multiple questions and identifying common mathematical patterns in those answers.
This mathematical approach groups individuals based on their underlying beliefs rather than superficial characteristics such as age, gender, or income. Using this method, the authors categorized the general public into four different types of belief systems regarding artificial intelligence. Researchers found that a segment of the public remains strongly supportive of artificial intelligence. These supportive people tend to believe that technology will bring about overwhelmingly positive changes in the economy.
They are likely to assume that technological tools will support human workers rather than completely remove them from the labor market. The group also appears to believe that companies can use this technology to benefit everyday consumers through lower prices and better products. On the one hand, this data provides evidence that a large portion of the population views artificial intelligence as a major societal threat. These people theorize that this technology will cause widespread harm to consumers.
They tend to think that companies replace human skills in the workplace with automated systems simply to save costs. For this threatened group, this technology represents a direct risk to job security and personal financial security. The authors found that these different sets of beliefs strongly predicted the kinds of government policies a person would support. People who see technology as a threat favor policies aimed at slowing or stopping job losses.
These restrictive policies may include strict regulations for businesses or complete bans on certain automated systems. The main purpose of such policies is to protect existing jobs exactly as they are now. People with more optimistic views about technology tend to support a very different set of policies. These people support government programs that help workers adapt to economic changes.
Adaptation policies could include funding for higher education, specialized job retraining programs, and funding for workers in new industries. The group acknowledges that some jobs will disappear, but believes society should focus on preparing the workforce for new opportunities. The researchers noted that these differing opinions on artificial intelligence closely align with existing political preferences. Voters are already starting to sort themselves into opposing camps based on broader political beliefs.
This data suggests that public attitudes toward technology are becoming polarized along standard partisan lines. This means that affiliation with a particular political party tends to predict that party’s position on technology regulation. The alignment of these beliefs and political parties is not completely random. Political parties often have established platforms on trade unions, corporate regulation, and free markets.
When voters apply these existing frameworks to new technical issues, they tend to adopt the perspectives of their elected political leaders. This dynamic suggests that artificial intelligence will be absorbed into current political debates rather than creating a new political spectrum. The authors concluded that fissures already exist in public opinion regarding artificial intelligence. These sectors provide easy opportunities for political entrepreneurs to gain support.
A political entrepreneur is a politician or activist who finds new issues to champion to attract voters. These numbers can mobilize various groups for their own political interests by capitalizing on people’s fears about unemployment and hopes for economic growth. One potential misconception of this study is the assumption that public opinion toward this technology is permanently fixed. Readers may think that these early political divisions persist over time.
This study captures a specific moment in time, meaning attitudes can easily change as technology becomes a more common part of everyday life. Personal, hands-on experience with automated tools may allay some fears and validate others. A limitation of this study is that it focuses only on the United States and Canada. These two countries share similar economic structures, which can influence how their citizens view labor and technology.
Public opinion may look very different in countries with stronger social safety nets or very different labor laws. Expanding the study to other regions of the world would help clarify how national culture influences these causal beliefs. Future research could track these public opinions over several years to see how attitudes change. Tracking the same group of people over time would provide evidence of how real-world economic changes affect individual beliefs.
Additional research could also look at specific professions to see if healthcare workers view technology differently than manufacturing workers. Exploring these details will further deepen our understanding of the emerging political landscape.
The study, “Causal Beliefs and the Potential for Political Backlash Against AI,” was authored by Sophie Bowin, Beatriz Magistro, R. Michael Alvarez, Bart Bonikowski, and Peter J. Loewen.
