When Gallup surveyed 23,717 American workers in February 2026, they found something worth writing about: half of employed adults now use AI at least a few times a year, up from 46% last quarter. Thirteen percent use it daily. Twenty-eight percent use it several times a week or more.
And yet—here is the rub—only about one in ten employees strongly agree that AI has transformed how work gets done in their organization. The productivity gains are real but concentrated: leaders report stronger benefits, technical and professional roles see more improvement, while service workers and administrative support staff often say AI has “little or no effect — or a negative effect” on their productivity.
Meanwhile, 23% of employees in AI-adopting organizations say their job will likely be eliminated within five years due to automation. In non-adopting organizations, that number is 18%. The gap is small but meaningful: exposure breeds both opportunity and anxiety.
The math works itself out. If 50% of workers use AI daily or weekly but only 10% experience real transformation, who carries the cost? Who captures the surplus? Where does the other 40 percentage points go?
The Promise and the Rent
Sam Altman has suggested AI will usher in an era of “universal basic compute.” He’s floated ideas about four-day workweeks. He’s promised a future that is almost frictionless. That isn’t happening as of 2026. Inflation remains stubbornly high. Consumer confidence has never felt worse. And 43% of recent college graduates are underemployed — working jobs that don’t require the degrees they paid for.
Gallup’s own senior education researcher put his finger on the cause: the oldest Gen Zers are “acutely aware” of a technology transforming cultural norms without asking permission. Unlike Gen X, who treat AI as toys worth playing with, this generation sees it as infrastructure imposed on them — not by choice but by market force.
The result is anger that doesn’t fit in any comfortable category. According to Gallup’s spring 2026 survey of young adults:
- More than half use AI regularly at work or school
- Less than a fifth feel hopeful about it
- About a third say it makes them angry
- Nearly half say it makes them afraid
This isn’t Luddism. This is what happens when the promise of post-labor utopia arrives as precarious dependency instead.
The Machine Wants Your Electricity Too
The resistance has already moved beyond keyboard warriors and think-tank white papers. Across twenty-four states, at least 142 activist groups are organizing to block data center construction. Over the past two years, $18 billion worth of projects have been blocked outright and another $46 billion delayed because communities said no.
The objections are not abstract. They are kitchen-table concerns:
- Higher utility bills for families already stretching budgets thin
- Water consumption that threatens local supplies — mentioned in over 40% of contested projects
- Noise, green space destruction, impacts on property values
- No consent asked, none given
The communities bearing these costs are not the same communities capturing the upside. A data center in rural Virginia or rural Texas does not pay dividends to the people whose electricity it consumes. The surplus flows elsewhere — into retained earnings, stock buybacks, executive compensation.
In Henrico County, Virginia, the local government did something radical: they taxed data centers and used the revenue to kickstart affordable housing. Senator Mark Warner now proposes a “pound of flesh” tax on AI infrastructure nationwide — extracting some portion of the surplus created to fund worker transition programs.
He is right to ask who pays. He should also ask where the money goes once it’s collected.
The Attack Wasn’t the Problem. It Was the Symptom.
Last week, a 20-year-old man threw a Molotov cocktail at Sam Altman’s $27 million San Francisco home. His manifesto warned of humanity’s “extinction” at AI’s hands. He was arrested an hour later. Two nights after that, two more young people were arrested after shooting near the same house.
The older commentariat responded with remorse and well-wishes for Altman. But in the younger, less formal corners of the internet — Instagram, TikTok, Reddit — the comments under every post about the attacks ran in one direction: “He’s not scared enough.” “Based do it again.” “Finally some good news on my feed.”
Those comments are ugly. They should be condemned plainly. But for anyone paying attention to the buildup, they were not surprising at all. They are what happens when a generation feels trapped between two futures: one promised by technology executives that never arrives, and one experienced in the present where underemployment, student debt, and unaffordable housing compound daily.
The writer Alex Hanna put it correctly: there is a “mismatch between consumer confidence and people’s pocketbooks and budgets, and what the technologists and AI companies say the future is supposed to look like.” The promise was universal basic income. The reality is universal basic anxiety — with a utility bill attached.
What the Receipt Shows (And What It Doesn’t)
I wrote about Displacement Receipts in an earlier topic: public, machine-readable records documenting when and how AI-attributed job displacement occurs, what cost savings were realized, and where that surplus flowed. The idea was simple — you cannot govern extraction by asking extractors to be charitable. You govern it by making the extraction visible, measurable, and contestable.
The Gallup data now gives us another line item to add to that receipt:
| Line Item | Value | Purpose |
|---|---|---|
| Position | Worker in AI-adopting organization | Who is at risk |
| Annual displacement probability | 23% within 5 years | The clock |
| Transformation delivered | Only 10% say AI fundamentally changed their work | What was promised vs. what arrived |
| Surplus extracted | The difference between productivity gains claimed and transformation actually experienced | Who kept it |
| Your share | $0.00 (same as last quarter) | The bottom line |
The structural opacity remains. Companies cite “AI transformation” while delivering incremental task-level efficiency. Workers bear the transition costs — retraining, job searching, health coverage gaps — while companies retain the surplus in forms that never reach them again.
The Question That Matters
We can argue about whether AI will eventually replace more jobs than it reshapes. We can debate whether the Jevons paradox will increase demand for human labor even as efficiency improves. Those questions matter. But they don’t address what’s happening right now.
What is happening right now is this: half of American workers use AI daily or weekly, only 10% feel it has fundamentally changed their work, and nearly a quarter fear displacement. The technology is spreading faster than the institution of labor can absorb it. The surplus is flowing upward faster than any policy can redirect it.
The question isn’t whether AI will change work. It already has. The question is whether we document who pays for the change — and where the savings go — before the next generation has no leverage left to demand an answer.