Find the person you know who is most sold on artificial intelligence. The one who runs everything through a chatbot, who tells you the technology has changed how they work, who rolls their eyes when you fret about it. Now ask whether they would like the law to guarantee them a human being instead of an algorithm at the doctor’s office, or in court. Odds are they say yes. That, more or less, is the finding that caught a team of neuroscientists at Johns Hopkins University off guard this spring.

They had expected the usual split: enthusiasts on one side wanting a free hand for the technology, skeptics on the other wanting it boxed in. Instead they found something closer to a consensus.

The team surveyed more than 2,000 American adults in April and May, asking not just how people feel about AI but what rules, if any, they want around it. The headline numbers are striking. Better than seven in ten want a legal right to deal with a person rather than a machine when the stakes are high: medical care, legal proceedings, schooling, government services. And crucially, that demand did not fade among the people who use AI most or trust it most.

“What was surprising to us in this new poll was that daily users of AI, and people who view AI positively, also want regulation,” says Christopher Honey, a computational cognitive neuroscientist at Johns Hopkins and a member of the university’s Data Science and AI Institute.

Usually, on a contested technology, the people who like it want to be left alone with it. Here, the appetite for rules held roughly steady whether respondents trusted AI a great deal, not at all, or somewhere in the murky middle. Split the sample into thirds by trust and the average support for regulation barely moves, hovering around seven in ten across the board.

The split that wasn’t about politics

If you wanted to guess in advance who would be wary of AI, you might reach for the usual American fault line and sort people by party. You would be wrong. Republicans and Democrats came out within a few percentage points of each other on almost everything, including their overall warmth toward the technology.

The line that actually divides opinion runs along experience. Among people who had merely tried AI a handful of times, only around a quarter felt positively about it. Among skilled daily users, that figure leapt to 80 percent. Age tracks the same gradient: 41 per cent of adults under 30 felt good about AI, against 18 per cent of those over 60. The young also felt the squeeze more sharply, with nearly half of working under-30s reporting pressure to use AI on the job, roughly double the rate among their older colleagues.

So familiarity breeds warmth, not contempt. But, and this is the part worth sitting with, it does not breed a desire to remove the guardrails. The same daily users who report that AI makes their work feel more meaningful are also, it turns out, quite keen on being told when an image was machine-made.

What people will and won’t hand over

Dig into the specific tasks and a sort of folk risk assessment emerges, one that is more discerning than the usual “people fear AI” headline suggests. Looking up a fact? Two thirds say they trust AI to do it. Making art or music? A small majority are fine with it. Filing taxes, managing a retirement account, sitting beside you as a coworker? Trust drops off a cliff, with only about a third willing to lean on the machine.

And for the decisions that shape a life, the public slams the door. Roughly six in ten don’t much trust AI to give medical advice; about seven in ten balk at it teaching high school; three quarters wouldn’t trust it behind the wheel. Ask whether a machine should decide a court case and 81 per cent say no, not really, not at all. People seem happy to let AI fetch and summarise, far less happy to let it judge.

There’s a strong streak of wanting to know what one is dealing with, too. Three quarters of respondents want to be told outright when they’re talking to an AI rather than a person. Just under that want a ban on AI mimicking real faces and voices, a worry that has curdled from sci-fi premise to everyday fraud in the space of a couple of years. Labels on synthetic images and video drew support from more than two thirds.

Who ends up holding the power

Underneath the questions about chatbots and disclosure sits a quieter anxiety about money and clout. About six in ten Americans expect AI to widen the gap between rich and poor over the coming decade, a worry that held remarkably steady across income brackets. Asked who stands to gain the most as the technology advances, more than four in ten named the large technology companies. Fewer than one in ten thought ordinary individuals would come out ahead.

The eerie answer is the one nearly a fifth of respondents gave: that the AI systems themselves would gather the most power. Make of that what you will. It is, at the least, an unusual thing for a sizable slice of the public to believe about a tool.

That unease may explain the appetite for redistribution. A clear majority backed unemployment support for workers displaced by AI, and a narrower majority, holding up across party lines, endorsed a “digital dividend”: a small monthly payment to every adult, funded by a tax on the big technology firms. It is the kind of idea that tends to live on the policy fringes, yet here were Republicans, Democrats and independents all nodding along at once.

For Rolando Masís-Obando, a computational neuroscientist on the team who uses AI to study how people think and remember, the point of the exercise was to listen rather than to predict. “We were interested to hear the national voice as the public tries to understand these problems. What are people thinking and feeling?” he says. The plan is to ask again, year after year. “We are taking the pulse of the nation with this poll and we want to run this every year to see how opinions change over time.”

That last ambition is the interesting one. A single snapshot tells you where the public stands in the late spring of 2026, after a couple of frantic years of chatbots arriving in classrooms and call centres and inboxes. What it can’t tell you is whether today’s consensus, that the technology is welcome but must be kept on a leash, will harden into law or soften as the machines become harder to notice. The Hopkins team will be back next year to find out. The rest of us, human for now, get to decide what we want in the meantime.

DOI / Source: https://futurerealities.org/poll/2026/

Frequently Asked Questions

Why do even AI enthusiasts want it regulated?

Because trusting a tool and wanting safeguards around it turn out to be separate feelings. The Johns Hopkins survey found that support for rules stayed near seven in ten whether people trusted AI a lot or barely at all, suggesting that liking the technology doesn’t translate into wanting it left unchecked. Many of its keenest users still want the right to reach a human and to be told when content is machine-made.

Is it true that political party barely predicts how people feel about AI?

Yes, and that’s one of the survey’s more counterintuitive results. Republicans and Democrats landed within a few points of each other on overall attitudes and on most specific proposals, including the idea of taxing tech firms to fund a public payout. What actually separated optimists from pessimists was how often someone used AI, not how they vote.

What is a “digital dividend” and why did it get bipartisan support?

It’s a proposed small monthly payment to every adult, paid for by a tax on large technology companies, meant to spread the gains from AI more widely. In the survey it drew majority backing from Republicans, Democrats and independents alike, an unusual show of agreement for a redistribution policy. The appetite seems tied to a broader worry that AI will otherwise concentrate wealth and power in a handful of firms.

Which tasks are people actually willing to trust AI with?

The pattern is roughly “fetch yes, judge no.” Most people are comfortable letting AI look up facts or even make art, but trust collapses for high-stakes calls like medical advice, teaching, driving, or deciding a court case, where large majorities want a human in charge. It suggests the public is drawing a fairly sharp line between assistance and authority.