Working-class Americans are worried that AI will eliminate jobs, including their own, and they want the government to step in with a massive retraining program for affected workers—funded in part by a tax on large corporations that replace workers with AI.
That’s one of several key takeaways from a new poll commissioned by Working Families Power, an organizing, advocacy, and research group that partners with the Working Families Party. The survey, released exclusively to The New Republic, also shows that people are suspicious of data centers, particularly their impact on energy costs, and they support temporary limits or even moratoriums on new centers. That isn’t so surprising given the protests against data centers around the country, but it’s yet another reminder of the huge opportunity for the Democrats in the rising populist rage against Big Tech.
So far, they’re largely missing it.
Working Families Power partnered with the Justice Research Fund to contact 2,511 working-class registered voters, oversampling in swing districts. To identify those voters, they used a rigorous definition of working-class that includes family income, education level, type of work, homeownership status, and whether someone has financial security to fall back on if they lose their job. It’s a far more exacting definition of working-class than simply sorting out voters without a college education, as many surveys and analyses do, and one that identifies voters who are suffering most from the affordability crisis.
The poll shows that 73 percent are “concerned” that AI will lead to job losses in the United States, while 62 percent worry it could “affect their own job, household income, or jobs of people close to them”; 58 percent think it will hurt working-class families overall. Among the jobs that a majority said AI would hurt are truck drivers; retail and service workers; office and administrative workers; and writers, designers, and other creative workers.
Huge majorities, 80 percent and more, want government support for retraining and education programs for those who lose jobs due to AI. They want regulations requiring corporations to give notice before replacing jobs with AI, and to be taxed to contribute to these programs. They also want to share in any benefits from AI through increased pay, shorter workweeks, and stronger benefits. Seventy-three percent agreed with this statement: “AI should work for working people, not just make billionaires richer while everyone else worries about losing their job.”
“What we’re seeing from results like this is that AI might be the biggest political realignment issue of our lifetime,” said Ravi Mangla, a spokesperson for Working Families Power and the Working Families Party. “It’s rare that you see numbers where there is 70, 80 percent agreement in a polarized country like this, and for the vast majority of people to agree that the government needs to step in and do something about AI job loss, that AI companies need to be paying for infrastructure and energy costs, and that there needs to be mass retraining, as well as benefits for workers who are laid off because of AI. That’s a level of agreement across politics that we don’t really see on anything.”
Mangla said Working Families Power doesn’t see enough candidates and officials addressing these widely held concerns. “[They] are not taking what is a political layup,” he said. “Neither party, right now, is rising to the occasion.”
Voters’ concerns extend to data centers. They’re worried that their growth would lead to increased utility costs, and 60 percent want a pause on construction to study their impacts. At the moment, only Vermont Senator Bernie Sanders and New York Representative Alexandria Ocasio-Cortez have released a comprehensive plan to address these concerns. The Congressional Progressive Caucus’s New Affordability Agenda also addresses issues like AI-driven surveillance pricing, and its chair, Texas Representative Greg Casar, supports taxing AI companies to pay for a jobs program. But the anger at Big Tech over AI is more expansive, and largely untapped on the campaign trail.
Professor Alondra Nelson, the founder and director of the Science, Technology, and Social Values Lab at the Institute for Advanced Study in Princeton, New Jersey, said these results confirm what many in her field have been arguing for a long time. “The costs of AI transformation should not be passed on to the public while profits flow exclusively to shareholders and executives. That’s not innovation—that’s extraction,” she said in an email.
Nelson said we need transparency first, because we’re still not sure how AI is affecting business and hiring practices. She said she’s encouraged by new tax policy ideas from the Center for Shared AI Prosperity, which include new taxes on wealthy people whose fortunes derive substantially from AI industries and requiring more contributions from AI firms to help affected workers.
The federal government is not pursuing such protections, and in fact the Trump administration and Republican-controlled Congress have tried to block states from regulating AI at all. “State legislatures are moving faster than Congress on some of these questions,” Nelson said. “That experimentation matters—which is precisely why we should resist any federal preemption of state-level AI regulation. Ultimately, though, we need federal standards establishing a floor of protection for working families, regardless of zip code.”
That’s especially true because the potential benefits of AI remain unclear, as tech evangelists simultaneously promise unimaginable abundance and warn of potential doom. “Even the tech CEOs and AI CEOs can’t tell you what’s going to happen,” Mangla said. “They will say this might destroy civilization or humanity. They’ll say outright that this will cause massive job loss. They are actually conceding to the massive risks and potential faults of this technology, and yet they’re resistant to commonsense policies that will actually protect people and that will allow people to continue to live their lives and earn a decent wage.”
A lot of the potential solutions, such as regulating companies like Anthropic and OpenAI, taxing them more, and providing a safety net for workers, are traditional progressive asks. But Mangla said the polled voters don’t see these as partisan solutions—just necessary ones. Working-class Americans don’t want to halt AI innovation, as the the poll shows, but they don’t want it to come solely at their expense.
