Why the AI Pessimism?

Americans fear worse jobs, little focus on everyday lives

AI Attitudes in Fall 2026

This first section of the report provides topline results from the Fall wave, reflecting the most recent data available.

Americans remain pessimistic about AI, with only 47% describing themselves as somewhat or very optimistic. Republicans are significantly more optimistic than Democrats, while all classes have negative to neutral views, with only the upper class reaching an even split.

Most Americans are unconcerned or enthusiastic about AI’s potential effect on their job, while those who are concerned are more likely to worry about worsening job quality than outright elimination. Perspectives are strikingly mixed by class, with the upper class most optimistic and the lower class most likely to anticipate no change. The middle class has the highest levels of concern. (Note, only respondents who held full- or part-time jobs were asked this question.)

The idea of a Universal Basic Dividend, explained in the survey as “a share of the profits earned by AI companies would be sent to all households to ensure that they benefit from AI’s growth, even if job prospects decline,” is broadly popular, with Democrats more supportive than Republicans.

But Americans do not want a Universal Basic Dividend as a substitute for work. Respondents were presented two possible futures with AI:

AI replaces nearly all human labor. The resulting wealth is shared among all people, who are free to spend their time however they choose.
AI improves the quality and productivity of jobs. People must still work to support themselves and their families, but the jobs are better and pay much higher wages.

By nearly four-to-one, they chose the latter.

Indeed, American views toward a Universal Basic Dividend are entirely uncorrelated with their preferred future. Regardless of their support for the dividend, they choose a world where “people must still work to support themselves and their families, but the jobs are better and pay much higher wages” by enormous margins.

Finally, the survey asked respondents for their perception of the priorities at the companies creating the most advanced AI systems, presenting ten different goals and asking for a response from “1 – not a priority” to “4 – a top priority” for each. Here, responses for each goal are shown as “net priority,” calculated as the share choosing 4 (“a top priority”) or 3, less the share choosing 2 or 1 (“not a priority”).

Americans see the AI labs focused on building the most powerful systems possible, earning high profits, and increasing user engagement, while giving little attention to labor market effects, preventing disastrous harm from their products, improving everyday lives, and especially protecting kids from inappropriate uses.

Shifts from Spring to Fall

While discussion of AI has dominated media and politics in recent months, American attitudes appear not to have shifted much or even in a consistent direction.

For example, optimism increased by 5 percentage points, while anticipated labor market effects did not change. Support for a Universal Basic Dividend declined by 4 percentage points, while preference for a future with high-quality jobs remained stable.

Perception of priorities for AI labs shifted lower for what had been (and still are) perceived as the highest priorities. Perception of solving major problems shifted toward a higher priority.

Drivers of AI Pessimism

The United States diverges from most other countries in its substantial pessimism about AI, which has major consequences for political decisions about the technology and creates obstacles to rapid expansion of infrastructure. While commentary generally assumes that opposition to AI stems from concern about job loss or catastrophe, results from this survey suggest the most important correlation is with beliefs about the priorities of the companies developing AI. The following data combines the Fall and Spring waves, for which responses were similar, to increase the statistical power of the analysis.

Unsurprisingly, expectations for labor-market impacts do correlate strongly with attitudes toward AI generally. Respondents who anticipate negative effects on their jobs tend to be much more pessimistic. What is surprising, though, is that it is not elimination of a job but rather worsening quality of a job that makes respondents most pessimistic.

Although concern about labor-market impacts is correlated with strong pessimism, not enough Americans share that concern for it to drive a substantial share of the nation’s pessimism. Recall, from the overall results, that only respondents holding a job were asked about effects on their job, and only 16% of them expect AI to affect their jobs negatively. They account for only 11% of the pessimists.

Differing perceptions of AI lab priorities are correlated with swings in optimism nearly as large as the job-related swings, but far more Americans are skeptical of those priorities than are concerned about labor market impacts. For five different goals, belief that the goal is a priority is associated with a swing of 52 to 72 points in net optimism, and in each case at least 39% of Americans believe it is not a priority.

Note that while perception of priority for socially valuable goals is closely correlated with general optimism, perception of high priority for commercial goals does not have a comparable effect in any direction.

Americans do not take issue with profit-seeking companies seeking profit. They are concerned, however, if they do not see a focus on broadly shared benefits as well.

About the Data

The American Compass AI Survey was conducted by YouGov in two waves. For the Spring wave, between May 1 and May 5, 2026, YouGov interviewed 1,039 respondents who were then matched down to a sample of 1,000 to produce the final dataset. The respondents were matched to a sampling frame on gender, age, race, and education. The sampling frame is a politically representative “modeled frame” of registered voters in the US, based upon the 2024 Current Population Survey (CPS) Voting and Registration supplements and the 2024 National Election Pool (NEP) exit poll.

For the Fall wave, between September 21 and September 28, 2026, YouGov interviewed 1,240 respondents who were screened for attention and data quality and then matched down to a sample of 1,000 to produce the final dataset. The respondents were matched to a sampling frame on gender, age, race, and education. The frame was constructed by stratified sampling from the 2024 Current Population Survey (CPS) sample with selection within strata by weighted sampling with replacements (using the person weights on the public use file).

In both waves, the matched cases were weighted to the sampling frame using propensity scores. The matched cases and the frame were combined and a logistic regression was estimated for inclusion in the frame. The propensity score function included age, gender, race/ethnicity, years of education, region, and home ownership. The propensity scores were grouped into deciles of the estimated propensity score in the frame and post-stratified according to these deciles.

The weights were then post-stratified on 2024 presidential vote choice as well as a four-way stratification of gender, age (4-categories), race (4-categories), and education (4-categories) to produce the final weight. The Spring wave was also post-stratified on a two-way stratification on race (4-categories) and education (4-categories).

In analysis of the results, “Class” is defined by education and income:

“Lower” (Spring N=155; Fall N=169): less than a four-year degree and household income below $30K; or did not report household income and do not have a high school diploma.
“Working” (N=281; N=286): less than a four-year degree and household income $30K–$80K; or did not report household income and have either a high school diploma or some college but no degree.
“Middle” (N=409; N=381): four-year degree or more and household income $30K–$80K; or household income $80K-$150K; or did not report household income and have a two-year or four-year college degree.
“Upper” (N=125; N=131): household income above $150K; or did not report household income and have a post-graduate degree.

Respondents with a four-year college degree or more but household income below $30K are excluded from analyses using the “Class” variable (N=30; N=33).