Marquette Poll Reveals 70% View AI as Harmful to Society

A new national survey from Marquette University Law School reveals something unusual in today's political landscape: widespread agreement across partisan lines. The poll, conducted April 8-16, 2026, found that 70 percent of respondents believe artificial intelligence is overall a bad thing for society. That's not a partisan split. It's not an age gap. It's a near-universal skepticism that cuts through the usual demographic fault lines.

The data comes from 982 adults interviewed nationwide, with a margin of error of +/-3.4 percentage points. Official documentation from Marquette Law School confirms the methodology and raw figures. What makes this finding notable isn't just the percentage—it's the consistency. Republicans, Democrats, independents, men, women, all age groups and income levels all landed on the same side of the question.

Poll Director Charles Franklin told WPR that the national results mirror a March poll of Wisconsin residents. "It really is striking," Franklin said. "I think what we're seeing is there's pretty much bipartisan skepticism, both here in Wisconsin and nationally." That's an awful lot of partisan agreement, where we normally see Republicans and Democrats on opposite ends of a question, not so in these national data. One thing that lies behind that is a really strong doubt about artificial intelligence.

The skepticism extends beyond AI to the physical infrastructure powering it. Majorities across all demographics believe the costs of data centers outweigh their potential benefits. This isn't abstract polling—it's playing out in real communities. Multiple data center projects around Wisconsin are facing public backlash. Some projects, like ones in DeForest and Menomonie, are even getting canceled or tabled amid public opposition.

Think about what that means for the industry. Data center developers have spent years selling a narrative of economic growth, job creation, and technological progress. The physical reality—transformers humming, cooling systems running 24/7, water consumption measured in millions of gallons daily—has apparently stopped resonating. (The industry's pitch deck probably didn't account for this level of pushback.)

WPR's reporting provides additional context on how these findings sit alongside other political sentiment. The Wisconsin Public Radio article details the broader scope of the survey, which covered tariffs, inflation, and Supreme Court approval. On tariffs, 66 percent of Americans agreed with a U.S. Supreme Court ruling overturning Trump's tariffs under the International Emergency Powers Act. That's higher than 59 percent of Wisconsin residents approving of the decision in the March Wisconsin poll.

But voter approval for that Supreme Court decision did not translate into overall approval for the nation's highest court. In the latest national survey, 57 percent of Americans said they disapprove of the way Supreme Court justices are doing their job. The last time a Marquette survey found a majority of people approving of the Supreme Court's actions was in May of last year with 53 percent approval. Marquette's Supreme Court poll hasn't found more than 54 percent approval for the court in any of its more than two dozen surveys since 2021.

The poll also asked Americans if they felt the country was a force for global stability or instability. For the first time, the Supreme Court poll included this question. The poll found that 60 percent of respondents found the country to be a global force of instability. Franklin noted that without historical data on this specific question, it's impossible to track the shift precisely. "But I can't believe that in earlier decades, there wouldn't have been a large majority who thought we were a stabilizing force in the world," he said.

What does this mean for AI developers and data center operators? The numbers suggest a credibility problem. When 70 percent of the population views your core technology as harmful, marketing campaigns about "innovation" and "efficiency" start to feel hollow. The physical friction of building these facilities—permits, community meetings, environmental reviews—has become a genuine bottleneck, not just a regulatory formality.

Franklin's observation about partisan agreement is worth sitting with. In an era where most polling shows Americans deeply divided on nearly every issue, finding consensus on AI skepticism is almost as surprising as the skepticism itself. The industry has assumed that economic benefits would eventually win over skeptics. That assumption may have been flawed from the start.

Whether this skepticism translates into policy restrictions remains to be seen. Local opposition has already killed some projects. Federal or state-level action would require sustained pressure. The poll captures a moment in time, not a permanent shift. But moments like this—when public sentiment aligns across partisan lines—often precede regulatory changes.

For now, the data centers keep building. The AI models keep training. The question isn't whether the technology works. It's whether the public will accept the costs of deploying it at scale. That's the real variable in the equation, and it's one the industry can't optimize away with better engineering.

Artūras Malašauskas is an AI Systems Integrator with 20+ years of production-grade web engineering experience. He has designed, shipped, and scaled enterprise Python/PHP systems for logistics, SaaS, and public-sector clients. For the past year, he has focused exclusively on AI integrations: deploying open-source LLMs, building generative media pipelines (image, audio, video), and engineering multi-agent workflows for real production environments. His standard: reproducibility, security, cost-efficient inference—no vaporware. He documents and evaluates emerging AI tooling, separating verified capabilities from marketing noise. Technical editor at: muza-ai.eu, ai-verslas.lt, ai-naujinos.lt Connect on LinkedIn