A recent survey indicates that while more Americans are utilizing artificial intelligence (AI), their comfort level decreases significantly when it comes to AI making decisions on their behalf. Currently, 72% of American adults report having tried at least one AI tool, compared to just 50% in November 2024, according to the National AI Opinion Monitor, which is operated by researchers at Rutgers University. Interestingly, over a quarter of people use AI daily, but fewer than 10% would let it make final decisions for them without any oversight.
Most Americans, it turns out, want greater government regulation of AI, although experts note that there is often a gap between the desire for regulation and the actual knowledge of how to implement it effectively.
“It’s not surprising that people prefer having a human involved in significant life decisions. It’s just human nature,” said James Czerniawski, who heads emerging technology policy at the Consumer Choice Center. He elaborated that AI systems can be designed to assist in decision-making while ultimately leaving the final choice to humans, which could encourage wider acceptance of AI technologies.
A survey led by #RUSCI Prof. K. Ognyanova finds Americans are embracing AI more than ever, but trust has limits. Fewer than 1 in 10 support AI making final decisions on hiring, loans, college admissions or parole without human oversight: https://t.co/Rrv0hlSpa5 pic.twitter.com/z9V75A2dym
— Rutgers School of Communication and Information (@RutgersCommInfo) September 9, 2026
The findings suggest a noticeable unease among Americans about allowing AI to make unsupervised decisions that can greatly impact their lives. Katherine Ognyanova, a Rutgers professor and coauthor of the report, stated, “People seem to prefer a role for AI, but not if it means letting it take the final decision.”.
Other analysts pointed out that such surveys often highlight the “trust gap” that comes before the widespread acceptance of new technologies.
Kristian Stout, director of innovation policy for the International Center for Law & Economics, noted parallels to early e-commerce, where consumers hesitated to input credit card information due to a lack of trust in online payment systems. Over time, as trust mechanisms improved—like better security protocols—adoption surged. Stout emphasized that the same would apply to AI technology; users need to trust that AI tools will function as expected.
Ognyanova added that while AI can be seen as useful, people are still wary of it undertaking critical tasks or making decisions without human input. “Using a tool is entirely different from having full trust in it,” she remarked.
According to the survey, a solid majority—59%—felt that AI should be regulated due to its associated risks, while 25% preferred limited regulation to encourage innovation. Stout observed that people often express a desire for free regulation but also want to continue using their favorite tools and receive transparency when AI evaluates them.
Some believe that negative narratives around AI have skewed public perception. Czerniawski commented that the portrayal of AI in media may mislead people about its actual capabilities. He warned that calling for strict regulation without understanding the technology could result in counterproductive regulations that harm consumer choices and global competitiveness.
The opinions on AI seem to divide along partisan lines, with significant differences in confidence levels. For example, in late 2024, 56% of Democrats felt AI served the public interest, whereas only 45% of Republicans agreed. Yet, recent figures from August 2026 showed a shift, with 56% of Republicans now expressing confidence in AI, compared to 44% of Democrats. Despite these shifts, 62% of Democrats and 57% of Republicans still want strict regulations.
The survey also revealed a concerning trend: many Americans may not fully understand AI, even if they believe they do. Although 59% said they comprehended how AI works, testing indicated that the average score was only 48% correct. Daily users scored just slightly better at 49%.
Ognyanova observed that confidence in AI grows much faster than actual knowledge about it. Stout added that people don’t need deep technical expertise to have reasonable opinions regarding liability and oversight around AI. Many expressions of a desire for tighter regulation may reflect a general concern rather than a constructive understanding of what that entails.
Moreover, more than three-quarters of respondents want to be informed when AI is used by employers to assess their performance. The survey gathered responses from 5,154 American adults between July 5 and August 1, conducted online by PureSpectrum, a market research technology firm. The report did not include information on the margin of error.
