A recent study published in Public Opinion Quarterly provides evidence that the public holds deeply divided beliefs about the economic impacts of artificial intelligence. The research suggests that these differing opinions tend to align with existing political divides in North America. The findings indicate that politicians could easily exploit these divisions, turning technology policy into a highly polarized political issue.
The research team included Sophie Borwein, Beatrice Magistro, R. Michael Alvarez, Bart Bonikowski, and Peter J. Loewen. These authors represent universities across the United States and Canada, including the University of British Columbia, Northeastern University, the California Institute of Technology, New York University, and Cornell University.
New technologies often produce overall economic gains for a country as a whole. At the same time, the specific costs and benefits of these advancements are rarely distributed equally among the population. This unequal distribution makes technological progress highly susceptible to becoming a polarized political talking point. Some groups might experience increased wealth and job opportunities, but other groups might face unemployment and wage decreases.
The authors wanted to understand the public’s causal beliefs regarding artificial intelligence. Causal beliefs refer to how people logically connect a specific cause to a specific effect in their minds. In this case, the researchers wanted to know how individuals think new software will directly impact different parts of the economy. They sought to map out who the public views 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 shift. Past technological revolutions, such as the rise of factory machinery, primarily affected physical labor. In contrast, modern automated systems are capable of performing complex cognitive tasks. This means that office workers, writers, and managers might feel a sense of economic vulnerability that they have not experienced during previous technological changes.
Knowing how people form these causal beliefs helps experts predict future political behavior. When individuals believe that a certain group is being unfairly harmed by a new policy or technology, they tend to support politicians who promise to stop that harm. The authors set out to determine if these attitudes about technology are already taking shape in the minds of voters. They predicted that these early beliefs would provide a foundation for future political movements.
To explore this topic, the researchers designed a novel survey instrument. A survey instrument is simply a structured questionnaire used to collect specific data from a large group of people. They distributed this 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 variety of viewpoints across different regions and economic backgrounds.
The survey asked respondents a series of questions about how artificial intelligence would affect various aspects of society. Questions covered everyday topics like consumer prices, product quality, and the overall job market. The researchers specifically asked whether the participants thought the technology would complement workers or replace them entirely. Complementing a worker means the technology acts as a tool that makes the person’s job easier or more efficient.
Replacing a worker means the technology does the job completely on its own, leaving the human worker unemployed. After collecting the thousands of survey responses, the researchers analyzed the data using a technique called latent class analysis. Latent class analysis is an advanced statistical method that helps researchers find hidden groups, or classes, within a massive set of complex data. It works by looking at how people answer multiple questions and identifying shared mathematical patterns among those answers.
This mathematical approach groups individuals together based on their underlying beliefs rather than surface traits like age, gender, or income. Using this method, the authors categorized the public into four distinct types of belief systems regarding artificial intelligence. The researchers found that a portion of the public remains highly supportive of artificial intelligence. These supportive individuals tend to believe that the technology will bring overwhelmingly positive changes to the economy.
They are likely to assume that technological tools will assist human workers rather than push them out of the labor market entirely. This group also seems to believe that businesses will use the technology to benefit everyday consumers through lower prices and better products. On the other hand, the data provides evidence that a significant portion of the public views artificial intelligence as a major societal threat. These individuals theorize that the technology will cause widespread harm to consumers.
They tend to believe that companies will use automated systems to replace human skills in the workplace simply to save money. For this threatened group, the technology represents a direct danger to job security and personal financial stability. The authors discovered that these distinct sets of beliefs strongly predict the types of government policies that a person will support. Individuals who view the technology as a threat prefer policies designed to delay or stop job loss.
These restrictive policies might include heavy regulations on businesses or outright bans on certain automated systems. The primary goal of such policies is to protect existing jobs exactly as they are today. People who hold more optimistic views of the technology tend to support an entirely different set of policies. These individuals favor government programs that help workers adapt to a changing economy.
Adaptation policies might include funding for higher education, specialized job retraining programs, or financial assistance for workers moving into new industries. This group accepts 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 align closely with existing political preferences. Voters are already sorting themselves into opposing camps based on their broader political beliefs.
The data suggests that public attitudes toward technology are becoming polarized along standard partisan lines. This means that a person’s affiliation with a specific political party tends to predict their stance on technological regulation. The alignment of these beliefs with political parties is not entirely random. Political parties often have established platforms regarding labor unions, corporate regulation, and free markets.
When voters apply these existing frameworks to new technological issues, they tend to adopt the viewpoint of their chosen political leaders. This dynamic suggests that artificial intelligence will not create a new political spectrum, but will instead be absorbed into current political debates. The authors concluded that cracks in public opinion regarding artificial intelligence already exist. These divisions provide an easy opportunity for political entrepreneurs to gain support.
A political entrepreneur is a politician or activist who finds new issues to champion in order to win over voters. By playing into the public’s fears about job loss or their hopes for economic growth, these figures can mobilize different groups for their own political gain. One potential misinterpretation of this study is the assumption that public opinion on this technology is permanently fixed. Readers might assume that these early political divisions cannot change over time.
The research captures a specific moment in time, meaning that attitudes could easily shift as the technology becomes a more common part of daily life. Personal, hands-on experience with automated tools might soften some fears or validate others. A limitation of the research is its focus solely on the United States and Canada. These two countries share similar economic structures, which might influence how their citizens view labor and technology.
Public opinion might look very different in countries with stronger social safety nets or highly different labor laws. Expanding the survey to other parts of the world would help clarify how national culture impacts these causal beliefs. Future research could track these public opinions over several years to observe how attitudes evolve. Tracking the same group of people over time would provide evidence of how real-world economic changes impact personal beliefs.
Additional studies might also look at specific professions to see if workers in healthcare view the technology differently than those in manufacturing. Exploring these details would add more depth to our understanding of the emerging political landscape.
The study, “Causal Beliefs and the Potential for Political Backlash Against AI,” was authored by Sophie Borwein, Beatrice Magistro, R. Michael Alvarez, Bart Bonikowski, and Peter J. Loewen.