This week's AI polling fits on one bumper sticker: executives have decided, no one else got a vote. Nearly every CEO surveyed now expects AI to cut headcount within two years. Their workers say they're more ready than the boss thinks. Their neighbors are rejecting the data centers that will power it all. Voters — even in deep-red primaries — want disclosure laws on AI campaign ads. And shoppers want a federal ban on AI-driven price tags before the technology hits the grocery shelf.
The constructive read, and there is one: this is not necessarily an anti-AI country. It's a country that has noticed adoption is being done to it rather than with it. That's a governance problem, which is the kind of problem we know how to fix.
Decided, Not Deliberating
Mercer's Global Talent Trends 2026 survey of 12,000 C-suite leaders, HR heads, investors, and employees across 16 countries lands on a number that doesn't leave much room for hedging. 99% of executives expect AI to reduce headcount in the next two years. 98% are planning organizational design changes in the same window. 63% name redesigning work around AI as the single biggest expected return on their people investments.
A number that high is partly an artifact of how the question was asked — "expect any headcount reduction" is a low bar that collapses a spectrum of conviction into a single near-unanimous figure. But even discounted, the direction is unmistakable, and the corroborating cuts below it (98% planning reorganizations, 63% redesigning work around AI) aren't soft.
And then comes the catch. Only 32% of those same executives believe their workforce can actually combine human and machine work effectively. So the plan is to restructure now, figure out the integration later. That sequencing — cut first, retrain second — is what's driving most of the rest of the polling this week.
A second Mercer cut on the same dataset gave us the worker-side number: the share of employees reporting they're thriving at work fell from 66% in 2024 to 44% in 2026. That's a 22-point collapse in two years — and Mercer attributes it specifically to anxiety about AI-driven displacement. Most anticipate these cuts will land on early-career roles, which is to say, future senior workers.
The Boss Is Wrong About the Workforce
The executive-level certainty doesn't extend to the executives' read on their own people.
A new Adecco Group survey of 2,000 C-suite leaders across 13 countries — covering 8.6 million workers — found a 31-point perception gap on AI readiness. 70% of workers say they're ready to collaborate with AI agents in their workflow. Only 39% of business leaders think their workers are ready. The gap is an infernal loop: a workforce that's more open than its bosses think, being managed by bosses who think the resistance is downstream when it's actually upstream — in the C-suite itself.
Adecco's own diagnosis lines up: only 31% of executives say leadership has enough AI knowledge to understand the risks and opportunities. The bottleneck on AI adoption isn't the cashier or the analyst. It's the floor above them.
Productivity gains from AI accrue across the org. The cleanup bill goes to the boss.
Though maybe there's good reason for the boss' reluctance: a Prolific survey of 2,057 U.S. workers found that 57% of managers say they've had to fix or redo work that a colleague over-relied on AI to produce — versus 38% of individual contributors. Among VPs and the C-suite, the share rises to 63%. The seniority gradient is the inverse of what you'd want: the most expensive labor in the company is now also doing the most quality-control labor.
The same survey found 44% of workers say their employer either has no clear AI policy or they don't know if one exists. 77% say they review AI-assisted work from colleagues more carefully than fully human-produced work. Only 17% say they always verify AI-generated information themselves. The companies racing to restructure around AI have, by their own employees' account, mostly skipped the governance step.
Not in My Backyard, Not on My Receipt
The corporate side of AI has a workforce problem. The civic side has a consent problem — and it's surfacing on two fronts at once: the land underneath the buildout, and the price tag at the end of it.
The data center revolt. In Utah, a Deseret News / Hinckley Institute poll found 53% of voters opposed to the Stratos Project, a 40,000-acre AI data center proposed for Box Elder County and backed by Canadian billionaire Kevin O'Leary. Just 30% support it. In a state that does not, as a rule, refuse business development, opposition is outpacing support by nearly 2-to-1.
That's not a one-state story, and the national pattern explains why. A Zencity survey of residents nationwide finds the abstract question — should we build more data centers — narrowly split: 35% in favor, 30% opposed, 33% unsure. But the moment the question gets concrete, the unsure column collapses. 86% of residents rate effects on local water supply as a decisive factor in whether they'd back a specific proposed project, and 85% say the same about electricity costs. Jobs (65%) and tax revenue (61%) come further down the list.
That's the dynamic playing out in Utah. Asked about data centers in the abstract, Americans split roughly down the middle. Asked about the 40,000-acre one being built down the road, they break 53–30 against. The undecided third has nowhere to go but opposition once a real project, with a real water bill, lands in their county. And the people most likely to be trusted in that fight aren't the local officials — only 1 in 5 residents say they'd believe a mayor's office on a project's impact. 57% would believe an independent environmental group instead.
The pricing revolt. A UFCW-commissioned poll by GBAO (Navigator Research) of 1,000 registered voters delivers the cleanest cross-partisan finding of the week. 67% of voters support banning AI-driven dynamic pricing and digital electronic shelf labels in grocery stores. The crosstabs show just how wide this support is: 67% of Democrats, 74% of independents, and 61% of Republicans all in.
72% of voters do not trust grocery stores to deploy these technologies responsibly. 58% say they'd be less likely to shop at a store that uses them. By more than 2-to-1, voters say they'd be more likely to support a candidate who backs the ban than less likely (41% vs 15%). Yes, the survey was commissioned by a union with skin in the game, and the question wording reflects that. But the cross-partisan magnitude is hard to write off.
The story across both fronts is the same. Americans are not turning against AI in general. They're turning against AI deployments that look like decisions made in a conference room they weren't in.
The Disclosure Consensus
The cleanest political opening for AI regulation this week came from the most unlikely place: an Oklahoma Republican primary poll.
NonDoc / Centiment surveyed 457 registered Republican voters ahead of the state's June 16 gubernatorial primary. The horse race is a genuine four-way tie. The AI findings are not. Nearly 90% of these Republican primary voters support requiring campaigns to disclose AI use in advertising materials. Roughly 75% support the state legislature banning the use of AI in campaign materials and ads entirely.
These are GOP primary voters — the segment of the electorate most resistant to new technology regulation by reputation, and the most receptive to "let the market decide" framing. They are also, on this particular question, lined up behind some of the most aggressive AI regulation any state has seriously considered.
That should send a signal to state legislators of both parties. AI campaign-ad transparency has the rare political profile of an issue where the activist base of the more skeptical party is ahead of the legislators. Movement here probably doesn't get punished. Inaction might.
This week Poll Vault tracked 13 polls touching AI and tech policy alone. See them all →
This Week's Shorter Stories
🧑⚕️ Caregivers love the tools, mostly. An Elevance Health Public Policy Institute survey of 2,041 unpaid caregivers of Medicaid beneficiaries found assistive technology positively impacted at least one aspect of wellbeing for 98% of users. 73% felt less anxious while caregiving; 67% felt less need to stay physically nearby. But 70% also reported technical errors, and 52% found upkeep burdensome. The opportunity for Medicaid managed care plans is wide; the gap between potential and product reliability is just as wide.
⛪ Christians are surprisingly open to AI for spiritual advice. A new Barna Group survey finds 54% of practicing Christians would trust AI for guidance on meaning and purpose; 48% for spiritual growth. At the same time, 83% worry about AI misrepresenting Scripture and 72% worry it could begin acting as a replacement for God. Among Millennial Christians, 44% consider AI spiritual advice as trustworthy as a pastor's — versus 34% of practicing Christians overall. Generational change is coming to the pew faster than most denominational leaders are tracking.
🇯🇵 The most surprising AI early adopter: elderly Japanese women. A Japan Institute for Promotion of Digital Economy and Community survey (n=1,449) found 48% of women aged 60–79 prefer AI over a human advisor for interpersonal relationship advice — the only demographic in the entire sample where AI beat humans, and a 23-point gap over elderly men (25%). The Chiba University researcher who studies AI and health called it surprising. The likely driver isn't tech savvy — it's the non-judgmental quality of the machine.
📱 Americans want more health tech, not less. A Reach3 Insights survey finds 71% of Americans now use a health app, and 88% want as much or more technology in their health and wellness over the next decade. The consent problem driving the rest of this issue doesn't show up here — when the technology is something people opt into for themselves, the appetite is enormous. That contrast is the whole story.
🔐 Every org cheats on its own security policy. A vendor-commissioned Replica Cyber survey of 200 U.S. cybersecurity leaders found 100% granted at least one formal security exception in the past year. 34% also handled exceptions through informal workarounds. 39% of organizations delayed or canceled a strategic initiative — including AI deployments — because the work couldn't be done securely. The C-suite is much more confident the environment is ready than the VP of Cybersecurity is: a 20-point gap on the same question.
Next Week
- Whether the surveillance-pricing ban becomes a real bill. UFCW's 67%-bipartisan finding is the kind of polling number that draws legislative oxygen. Watch for state AGs and consumer-protection caucuses to move first.
- Whether the Utah data-center pushback spreads. Box Elder County is a leading indicator. With 57% of Americans trusting environmental groups over local officials on these projects, expect the next 12 months of permitting fights to look more like pipeline politics than like industrial park politics.
- Whether 2027 early-career hiring contracts. Mercer's reporting that most AI-driven cuts will land on entry-level roles is the canary. If summer 2026 internship-to-offer conversion drops materially below 2025 baselines, the class of 2027 walks into the labor market the executives are designing on the whiteboard right now.
- Whether state legislatures move on AI campaign-ad disclosure before November. Oklahoma GOP primary voters just told their party they want both disclosure and outright bans. If a red state moves first, expect a wave of copycat bills.
What Else We Tracked This Week
- Most Workers and Students Have Avoided AI Over Ethics, Accuracy, or Environmental Fears — CNBC / SurveyMonkey
- Seven in Ten Americans Oppose AI Data Centers in Their Communities — Gallup
- Voters Trust Unions Far More Than Corporations, Politicians on AI in the Workplace — AFL-CIO / David Binder Research
- Voters See Institutions Declining and Reject Self-Driving Cars by Wide Margin — Echelon Insights
- Voters Nearly Split on Data Centers, Favor State Over Federal Regulation — RMG Research
- Americans Use AI Widely But Trust It Little, Fearing Surveillance and Corporate Motives — Howdy
- Patients Accept AI for Scheduling but Reject It for Billing and Diagnosis — Sogolytics
One More Thing
The temptation, reading this all stacked up, is the doom read: executives are barreling ahead, workers are anxious, communities are angry, and nobody trusts the people in charge of any of it.
The data actually supports a different read. The same week that 99% of executives told Mercer they're planning layoffs, 70% of workers told Adecco they're ready to work alongside AI. 98% of caregivers using assistive tech said it improved their wellbeing. 71% of Americans are using a health app, and 88% want as much or more technology in their health and wellness over the next decade. 48% of practicing Christians would trust AI for advice on spiritual growth.
Americans are not against AI. They are against AI deployments that skip them. The ban-the-surveillance-pricing finding, the data-center opposition, the Oklahoma campaign-ad disclosure consensus — these aren't anti-technology votes. They're pro-consent votes. The country isn't asking for AI to slow down. It's asking to be asked.
That's a fixable problem. The bosses got their vote. Everyone else is still waiting for theirs.
Until next time, Alex Lundry
Due Data is powered by Poll Vault. Get the full data behind every poll mentioned above.
Want the full data behind every poll mentioned above?
Create free account →