39% of Americans Expect AI to Do More Harm Than Good: Gallup

For AI developers, the latest Gallup data is a wake-up call: nearly 4 in 10 Americans now see artificial intelligence as a net negative, even as 70% report being knowledgeable about the technology. This growing dissonance between familiarity and comfort signals a communications failure that could slow enterprise adoption, stiffen regulatory headwinds, and poison the well for public investment in AI innovation. The industry must urgently close the gap between the 'inside baseball' advances in medicine and climate science and the frustrating, low-quality AI experiences that dominate everyday perception.

A new Gallup survey reveals a striking paradox in American attitudes toward artificial intelligence: while familiarity with the technology is climbing, comfort is plunging. In 2026, 70% of Americans say they are at least somewhat knowledgeable about AI, a notable jump from 64% in 2024. Yet over the same period, the share who believe AI will do more harm than good has surged to 39%, up from 31% each of the past two years. Only 9% expect AI to be a net positive force, while roughly half foresee an equal mix of good and bad. The data marks a significant deterioration in public confidence, especially among younger adults—47% of those aged 18 to 29 now predict more harm than good, a figure that far outpaces older cohorts and underscores a generational unease with the technology’s trajectory.

“

In 2026, 70% of Americans say they are at least somewhat knowledgeable about AI, a notable jump from 64% in 2024.

The findings, drawn from Gallup’s annual polling on technology and society, paint a picture of a public increasingly exposed to AI’s visible shortcomings: the proliferation of low-quality AI-generated content—often derided as “slop”—on social media, frustrating interactions with customer-service chatbots, headlines linking AI to job displacement, and growing scrutiny of the energy-intensive data centers powering models. Anton Dahbura, co-director of the Johns Hopkins Institute for Assured Autonomy, notes that this “partial familiarity” is toxic because people form sweeping negative judgments from narrow, often irritating encounters without ever engaging with AI’s less visible but transformative applications. He calls it a “period of instability with a disruptive new technology,” where each person sees a different facet and labels it the same generic term, missing the “inside baseball”—the behind-the-scenes use cases in drug discovery, materials science, aircraft design, and energy optimization that are quietly delivering real value.

For the AI industry, these sentiment numbers should set off alarm bells. Public opinion is a critical driver of regulatory momentum and enterprise adoption. When nearly 40% of the population distrusts the technology’s net impact, policymakers face growing pressure to impose guardrails—potentially stalling innovation or saddling startups with compliance burdens that only incumbents can bear. Already, the European Union’s AI Act and numerous U.S. state-level proposals reflect a desire to preempt harms. A further souring of public mood could accelerate such efforts, even as many beneficial applications remain in early development. Moreover, enterprise customers read the same headlines; a skeptical workforce may resist AI tools, slowing productivity gains that companies are banking on.