Polling as Propaganda: How Blue Rose Research’s AI Survey Misleads
Blue Rose Research’s new report, AI Is Colliding With America’s Affordability Crisis, presents itself as an objective window into American public opinion on artificial intelligence. It is not. If you read carefully, it becomes clear that it is a Democratic Party messaging document.
The last slide admits as much, revealing that the poll’s ultimate purpose is to identify which AI-related rhetoric most effectively “increases support for Democrats.” Everything upstream of that conclusion should be interpreted accordingly.
This is important because the progressive left appears to believe its best hope for electoral success in 2026 is to advance a narrative centered on inflation, job loss, and other disruptions, rather than a message about how Democrats can do a better job than Republicans at growing the economy.
The report’s most fundamental problem is the design of its questions. Throughout, respondents are presented with false and tendentious dichotomies.
On AI and worker protection, for instance, they are asked to choose between “providing help for American workers… even if that means limiting the amount that American tech companies can profit” versus “providing incentives for American tech companies… even if it allows tech companies to profit while eliminating jobs.”
This framing presupposes a zero-sum conflict between workers and tech companies, ignoring the substantial economic literature demonstrating that productivity-enhancing technology broadly raises living standards over time.
Unsurprisingly, when you construct a question to sound like a choice between helping workers and enriching corporations, most respondents choose the former. That tells us nothing about what people actually want from AI policy.
The poll also primes respondents with emotionally loaded language before measuring their views. The framing of an economy “already rigged for the elite” that uses “new technology to further stack the deck” against the American people is introduced not as a hypothesis to test but as an established backdrop. It then treats the resulting 64 percent agreement as confirmation. This is circularity dressed up as data.
Similarly, the poll conflates general cost-of-living anxiety with technology-related concerns—specifically, the rise of AI. It artificially manufactures the appearance of an AI-driven economic crisis when the underlying grievance is simply inflation and wage stagnation.
The report also presents the “cost-of-living crisis” as a settled reality, despite the fact that U.S. median wages have grown, albeit modestly, over the past year.
The selective presentation of data compounds the bias. The report buries the most balanced finding, that Americans are almost evenly split on AI’s future: 44 percent optimistic and 41 percent pessimistic. Meanwhile, it leads with the alarming figure that 69 percent of Americans believe a superintelligent AI would be “mostly harmful.”
The latter is a science-fiction hypothetical; the former reflects actual public sentiment. Choosing which number to headline is an editorial decision, not a scientific one.
Most egregiously, the report’s final section shows respondents a political ad script that claims, “Within 5 years, AI is projected to eliminate 75 percent of our jobs.” This figure has no credible empirical basis—mainstream economic research on AI and employment produces estimates nowhere near this magnitude.
Blue Rose then measures how effectively this false claim moves voters toward Democratic candidates. They are not measuring public opinion; they are optimizing disinformation. And more fundamentally, how does the so-called affordability crisis get solved without higher productivity, which necessarily means that technology eliminates some jobs and, in the process, lowers prices?
None of this means the underlying public anxieties about AI are manufactured. Economic insecurity is real, and concerns about technological displacement deserve serious policy attention.
But a poll designed from the outset to test Democratic messaging, built on leading questions and false choices, and culminating in an explicit partisan optimization exercise, is not a contribution to that serious conversation. It is a campaign tool wearing a white lab coat.
Rather than engage in this kind of anti-technology, anti-business, populist propaganda, the poll should have focused on the real issue: The frankly poor job that local, state, and federal governments do in helping workers who lose their jobs due to technology (or trade) reenter the labor market. That’s the question that matters, and one that ITIF has already addressed with a comprehensive agenda.
