Americans Are Using AI More Than Ever — And Trusting It Less

The standard model of technology adoption assumes a trust flywheel: people try a tool, see it work, and grow more confident over time. A new Quinnipiac University national poll, released Monday, suggests AI is running the flywheel in reverse.

The survey, conducted across a nationally representative adult sample, finds that AI tool usage has continued to climb — more Americans report regular use of AI assistants, writing tools, and AI-powered search than at any prior measurement point in the series. But confidence in AI accuracy has declined in parallel. The divergence is not a rounding error. It is a systematic pattern: usage up, trust down.

Familiarity Is Breeding Skepticism

The most striking finding is not the distrust itself — that has been documented in prior surveys — but its correlation with usage. Infrequent AI users express relatively higher trust. Heavy users express significantly lower trust. The pattern suggests that the primary driver of skepticism is not theoretical concern about AI but direct, lived experience with AI errors, hallucinations, and confident-sounding inaccuracies.

This is a meaningful reversal of how most technology trust curves work. People who use Google Search more tend to trust it more. People who use GPS navigation more develop stronger reliance on it. AI tools are generating the opposite dynamic: the more you use them, the more examples of failure accumulate in your experience, and the more calibrated your skepticism becomes.

Implications for AI Adoption Narratives

The finding complicates the dominant AI industry narrative that adoption automatically translates into trust, and that trust drives deeper integration. If usage and trust are decoupling, the adoption curve and the value capture curve may also be separating. Users who use AI tools frequently but trust them less are more likely to invest in verification workflows, maintain human oversight, and resist the kind of autonomous AI deployment that generates the highest margin for AI vendors.

For enterprise buyers evaluating agentic AI deployments — where autonomous action, not just text generation, is the value proposition — a workforce that is skeptical of AI accuracy is a structural adoption barrier that product improvements alone may not resolve. The trust deficit is not a perception problem. It is, at least in part, an accuracy problem wearing a perception problem's clothes.