AI's Promise vs. Public Trust: Why Americans Want Guardrails
We know artificial intelligence can unlock new possibilities and alleviate suffering. We also know we do not trust it to make decisions for us. A recent poll reveals a public caught between the promises of AI and the very real pitfalls of unregulated tech.
The latest Economist/YouGov poll of 1,604 U.S. adult citizens, conducted from May 29 to June 1, 2026, highlights this tension. While 28% expect AI to affect their lives in a somewhat positive way, 35% anticipate a negative impact. The apprehension only grows when we talk about the broader economy.
Economic Anxiety and the Tech Exodus
Americans are more pessimistic about AI's economic impact than its personal benefits, with 29% viewing the economic consequences more negatively than positive. College graduates feel this most acutely. The poll reports that 32% of college graduates are more negative than positive about AI's effect on the economy, compared to just 16% who are more positive.
This skepticism is not baseless. It comes amid widespread disruption in the job market, especially within the technology sector. In 2025, companies attributed 55,000 job cuts to AI, which is 12 times the number from just two years earlier. When tech companies shed workers at this pace, the public rightly questions who this technology actually serves.
Progress vs. Ethics: The Public Divide
Expectations about AI's broader impacts are not uniformly negative, though. A majority, 54%, agree that AI will change the workforce just as other technological tools have throughout history. New technologies often unsettle the labor market before expanding our productive capabilities. Americans may fear societal disruption, but they do not reject progress.
Across categories, more than 60% agree that AI will enable widespread progress we cannot yet anticipate. Yet trust in AI's ability to evaluate ethical dilemmas remains critically low. Among Democrats, Republicans, men and women, more than 70% agree that they do not trust AI to make ethical decisions.
Among Democrats, Republicans, men and women, more than 70% agree that they do not trust AI to make ethical decisions.
The Deregulation Dilemma
This deep distrust comes amid mounting concerns over AI regulation. In 2025, the Trump administration adopted America's AI Action Plan, which moved federal AI policy away from prescriptive directives in favor of fewer regulations. Critics have argued that additional guardrails are necessary to relieve Americans from tech monopolies that sacrifice the interests of everyday people.
On June 2, the Trump administration issued a new executive order that modestly increases oversight of cutting-edge AI models while preserving its broader deregulatory stance. The shift echoes the uncertainty surrounding AI governance. How do we preserve the innovations Americans expect from AI while addressing concerns that its risks require more targeted regulation?
The policy shift mirrors a broader public unease. The overwhelming plurality, 43%, concludes that AI cannot reason better than humans. Americans may accept AI as a personal tool, yet they remain wary of treating it as an independent actor that can make broadly impactful decisions.
Who Sets the Guardrails?
Despite regulatory uncertainty, many Americans embrace AI for its potential. A majority, 61%, believe AI has a strong capacity to alleviate suffering and unlock new possibilities. This view is widely bipartisan, with 65% of both Harris and Trump supporters holding it.
As AI becomes embedded in personal and professional life, it creates tension. A lighter regulatory approach enables greater innovation, expanding AI's capacity to improve life, health, and workplace efficiency. Yet that framework leaves Americans uncertain about who will set the guardrails on this rapidly advancing technology.
We cannot rely on tech monopolies to police themselves. If we want to harness AI's promise without sacrificing our values, we need robust, democratic oversight. The future of AI must be shaped by the public, not just the powerful few.