Most Americans believe that artificial intelligence is advancing too fast. A survey released on October 8 by the AP-NORC Center for Public Affairs Research revealed that 64% of adults in the United States consider the speed of the technology's development excessive, while 27% rate the pace as adequate and only 8% believe the evolution should be faster.
The concern cuts across political divisions, although it is more intense among Democrats. In this group, 76% consider the advance too fast, a share that falls to 54% among Republicans.
The survey also identified a difference between what Americans expect from the government's actions and the trust placed in those responsible for conducting artificial intelligence policy.
Human control and job protection outrank the technology race
When asked about the priorities of the American government, 81% of respondents considered it extremely or very important to keep artificial intelligence under human control. Another 77% attributed the same importance to protecting American workers.
These concerns were given more prominence than goals associated with technological expansion. Promoting innovation was considered highly important by 51% of participants, while only 45% gave that rating to competition with other countries, such as China.
The contrast also appears in the assessment of the political class. 39% of Americans trust neither Republicans nor Democrats to adequately handle issues related to AI. Another 23% trust the Democratic Party more, 21% prefer the Republican Party and 16% consider both equally capable.
President Donald Trump also faces resistance in this area. Only 31% approve of his handling of issues related to artificial intelligence, while 67% disapprove.
The survey also examined the use of AI tools in the U.S. legislative elections in November. About 21% of respondents said they had already used or intended to use the technology to research candidates or issues on the ballots.
A smaller group, equivalent to 10% of adults, said they had already turned to or planned to turn to AI to help decide who to vote for. The questionnaire combined actual use and future intention in a single response, without distinguishing how many participants had already used these tools for that purpose.
Conducted between September 24 and 28, 2026, the survey interviewed 2,140 American adults through online and telephone interviews. The overall margin of error is 2.9 points, plus or minus.
