(DCNF)—Americans are increasingly using artificial intelligence (AI), but they become less comfortable when it makes decisions for them, according to a new survey.
Seventy-two percent of American adults have used at least one AI product, up from 50% from November 2024, according to the National AI Opinion Monitor, a project led by Rutgers University researchers that tracks public opinion on AI. More than a quarter of Americans use AI every day, although a less than one-in-ten would allow it to make final decisions for them without supervision.
More than a majority of Americans want stricter government oversight of AI, although experts believe there is a big difference between wanting regulation and knowing how to regulate AI.
“Americans wanting a human in the loop for major life decisions isn’t all that surprising. In fact, it’s human nature. People naturally are going to want agency over high-stakes choices like hiring, loans, or medical care,” James Czerniawski, the head of emerging technology policy for the Consumer Choice Center, told the DCNF.
“AI agents are intentionally malleable by design, and can be easily modified to deploy in an approval-based architecture, where it does the heavy lifting, data processing and prep work and turns the actual decision making back to the human for the final nod,” Czerniawski added. “If it is used in such a fashion, one could easily see AI agent adoption accelerate.”
The survey found that Americans become far less comfortable having an AI model make the final decision without them.
“People seem to draw the line at letting AI make unsupervised decisions that significantly affect human lives,” Katherine Ognyanova, a professor with the Rutgers School of Communication and Information and a coauthor of the report, said in a statement. “People are somewhat comfortable with a role for AI, just not letting it make the final call.”
Other experts said that the poll showcases that there if often a “trust gap” that precedes technology adoption.
“Early e-commerce had the same problem. Consumers would not put credit cards into websites because the payment infrastructure was not trustworthy. Adoption followed the build-out of the necessary trust infrastructure. This includes Amazon normalizing the “paying online” experience, SSL and secure payments becoming well known ways for consumers to trust sites, and credit-card networks and issuers building anti-fraud technology and policies into the system. Once the risk in the system felt manageable, consumers adopted in droves,” Kristian Stout, the director of innovation policy for the International Center for Law & Economics, told the Daily Caller News Foundation in a written statement.
“The market cannot broadly adopt them until counterparties and the people deploying them can trust what they will do,” Stout remarked, adding that the “mechanisms that get us over this hump are mostly institutional, not psychological.”
“We adopt AI because it is useful, convenient or expected of us,” Ognyanova said. “At the same time, people are concerned about AI taking over high-stakes tasks or making important decisions without human supervision. Overall, using a tool and trusting it fully are two different things.”
“As those guardrails mature, expect the same adoption curve e-commerce followed,” Stout said.
Most Americans want more government oversight of AI, according to the Rutgers survey.
Fifty-nine percent believe that governments should regulate AI due to its potential risks, compared to 25% who want limited regulation to foster innovation, the survey found.
Consumers often say they want strong regulation “in the abstract,” but the conversation becomes more nuanced when pushed for details, Stout remarked.
“People say regulate AI in a general sense. They also say they want to keep using the tools they love. They want disclosure when AI evaluates them. They want the systems that help them to thrive,” Stout explained. “This is very similar to the gaps we see in privacy surveys and what technologies consumers actually want to use every day. The gap between strict regulation and limited technical understanding is partly an artifact of survey design. A blunt preference question and a fine-grained knowledge quiz measure different things.”
Others believe that a wave of negative press has given Americans a false impression about AI.
“People have been force-fed an image of AI shaped by Hollywood and a relentless wave of hyped-up AI ‘doomerism’ on their feeds rather than what the technology actually is. When even daily users fail a basic quiz on how these tools work, but are simultaneously calling for ‘strict regulation,’ that is a massive red flag from a policy standpoint,” Czerniawski said.
“Demanding strict guardrails around technology you don’t even understand is a recipe for drawing up bad regulations that will ultimately stifle consumer choice, inhibit competition, and run the risk of undermining the country’s status as a global leader in emerging technology to our adversaries like China, whose vision for this technology is inherently anti-American,” he added.
This belief translates to partisan lines.
In late 2024, 56% of Democrats believed AI can act in the public interest, compared with 45% of Republicans. In August 2026, 44% of Democrats expressed confidence in AI, whereas 56% of Republicans said they believed in AI to act in the common good. Overall confidence remains at 46% over the last two years.
“While overall confidence barely moved, the two groups have switched places. Notably, both parties seem to agree on policy preference: 62% of Democrats and 57% of Republicans still want AI strictly regulated,” Ognyanova said.
The survey found that it appears that Americans may not know as much about AI as they think they do.
Fifty-nine percent of Americans said they understand how AI works; however, when researchers quizzed those polled about how AI works, the average score was 48% correct. Daily AI users received a 49% correct score when tested for AI familiarity.
“Use appears to increase confidence considerably faster than it increases knowledge,” Ognyanova remarked.
“People do not need deep technical understanding of a system to form a reasoned view about who should be accountable for its harms. A sound preference about oversight and liability does not require knowing the architecture underneath. And, so an expression of ‘AI should be more regulated’ is probably more like a general vibe for the survey takers, not a really well thought out set of mandates of what ‘strictly regulated’ means in practice,” Stout said.
More than three-quarters of Americans want to know when employers use AI to evaluate their performance.
Researchers polled 5,154 American adults between July 5 and August 1. PureSpectrum, a market research technology company, conducted the surveys online. The report does not disclose the poll’s margin of error.
All content created by the Daily Caller News Foundation, an independent and nonpartisan newswire service, is available without charge to any legitimate news publisher that can provide a large audience. All republished articles must include our logo, our reporter’s byline and their DCNF affiliation. For any questions about our guidelines or partnering with us, please contact [email protected].
