Gallup and the Special Competitive Studies Project just published their second annual read on how Americans feel about artificial intelligence, and the headline finding should stop every executive rolling out AI tools this quarter. Use is up. Knowledge is up. Trust is not. In fact, optimism about what AI will do for the economy has reversed entirely in twelve months, and the reversal is sharpest among the younger workers most organizations assumed would be their easiest adopters.
This is not a story about the general public. It is a preview of what happens inside any organization that treats AI adoption as a tool rollout instead of a trust problem. Leaders at nonprofits, associations, healthcare systems, and professional services firms have spent the past year buying licenses and running pilots, then wondering why usage never turns into the productivity gains the vendor promised. The Gallup data explains why, in numbers pulled directly from a nationally representative survey of more than 3,000 U.S. adults. What follows is what the report actually found, and what it means for anyone trying to get a workforce to actually use AI rather than merely have access to it.
Familiarity Rose. Trust Did Not Move With It
Reported AI use among U.S. adults climbed from 39 percent in 2025 to 51 percent in 2026, a 12 point jump in a single year. Self-reported knowledge about AI rose alongside it, from 60 percent to 68 percent of adults describing themselves as very or somewhat knowledgeable. By the usual logic of adoption curves, trust should have followed. It did not. Only 28 percent of Americans fully or somewhat trust AI to make fair and unbiased decisions in 2026, essentially flat against 32 percent in 2025, and more than twice as many people distrust AI as trust it.
The lesson for any leader tracking adoption metrics is direct. Login counts, seat utilization, and self-reported comfort with a tool are not proxies for trust, and treating them as interchangeable is how a rollout looks successful on a dashboard while quietly failing in practice. If your organization is measuring AI adoption purely by usage, you are measuring the same thing Gallup found rising nationally while trust sat still. Build a separate measure for trust and confidence, gathered directly from the people using the tool, and track it alongside usage rather than assuming one implies the other.
Regular Users Trust It. Everyone Else Is Losing Ground
The report's clearest predictor of trust is simple: people who use AI regularly trust it at more than three times the rate of people who rarely or never use it, 43 percent versus 12 percent. What makes this finding urgent rather than reassuring is the direction of change. Trust among infrequent or non-users roughly halved over the past year, falling from 23 percent to 12 percent, while trust among regular users held steady. The gap between your adopters and your holdouts is not closing on its own. It is widening.
Inside an organization, this maps directly onto the employees who skipped the initial rollout, whether from skepticism, workload, or simply not being in the first wave of training. Leaving them to catch up informally is not a neutral choice. The data suggests their trust erodes further the longer they stay outside the loop, which makes the eventual conversion harder, not easier. A phased rollout without a deliberate plan to bring late adopters in produces two workforces: one that trusts the tool because it uses it daily, and one growing more skeptical by the month. Closing that gap requires scheduled onboarding for holdouts specifically, not an open invitation to join whenever they are ready.
Economic Optimism Reversed, and Younger Workers Felt It Hardest
A year ago, more than twice as many Americans expected AI to boost the economy as expected it to hurt it, 53 percent to 24 percent. That has flipped. In 2026, 45 percent expect AI to reduce economic growth against 36 percent who expect it to boost growth. Fifty-nine percent now expect AI to destroy more businesses than it creates over the next decade, and 70 percent expect it to destroy more jobs than it creates, both figures roughly 10 points worse than a year earlier. Only one in five Americans believe most displaced workers will be able to retrain into a different kind of work.
The age pattern matters most for employers. Adults aged 18 to 29, the group organizations often assume will champion new tools, are the most pessimistic of any age cohort: 57 percent think AI will reduce economic growth against 26 percent who think it will boost it. Adults 65 and older are the only age group where more people expect AI to help growth than hurt it. If your rollout plan assumes younger staff will carry enthusiasm for the organization, this data says the opposite is more likely true right now, and a rollout built on that assumption will meet resistance it did not anticipate. Address the concern directly in training and communication rather than assuming it away.
What Actually Builds Trust: Transparency and Independent Testing, Not Speed
Gallup asked directly what would increase trust in AI, and no single answer dominated, but three converged around similar strength: greater transparency from companies about how AI makes decisions (49 percent), testing by independent experts (48 percent), and testing systems for fairness before public deployment (46 percent). On the broader question of regulation, the public is far more aligned than the specific policy fights suggest. Seventy-nine percent say government should prioritize safety and data security rules even if that slows AI development, against just 7 percent who prioritize speed. Seventy-five percent believe safety testing should be conducted by independent experts rather than by the companies building the tools.
This is a governance blueprint any organization can apply internally without waiting on federal policy. Publish, in plain language, what a given AI tool is and is not making decisions about inside your organization. Route any AI system that touches hiring, patient information, member data, or client-facing decisions through review by someone outside the team that built or purchased it. Test for fairness and error before wide deployment, not after a complaint. These three moves mirror exactly what the public says would rebuild its own trust in AI, and there is no reason the same logic stops working at the door of your organization. This is precisely the oversight structure we build with clients through the AI Adoption Advisory work at Breakfast Leadership Network, because the tool was never the hard part. Deciding what it is allowed to decide, and proving that to the people affected, is.
Training Is Happening Informally. Formalize It Before Your Competitors Do
More than twice as many Americans have asked an AI for help using AI as have asked another human, 57 percent versus 23 percent, meaning most people are teaching themselves in real time with no structure and no one checking their work. Formal channels lag well behind: 26 percent have used online training videos or articles, 16 percent completed employer-provided training, and only 9 percent have taken a course or certification program. Yet when asked about formal investment, support is strong and notably bipartisan: 72 percent support promoting STEM education to prepare workers for AI-driven jobs, and 66 percent support on-the-job training programs specifically.
That gap between informal self-teaching and desired formal training is where most mid-sized organizations are quietly failing right now. Handing someone a license and a Slack channel is not training, and the Gallup numbers suggest most employees know it. Building a structured, role-specific training path, with clear decision rights for what the AI tool handles versus what a human is accountable for, is the difference between an activation number that looks good in a board report and a workforce that actually trusts and uses the system day to day. This is the exact structural gap our Leadership Operating System work is built to close, because a tool rollout without a decision framework behind it is a pilot program that never graduates.
The Leadership Takeaway
The public's relationship with AI in 2026 is not a story of resistance to new technology. It is a story of rising exposure without rising confidence, and every organization deploying AI right now is running the same experiment on a smaller scale. The fix Gallup's own data points to is not more tools or faster deployment. It is transparency about what the system decides, independent review before it touches anything consequential, and formal training that replaces the guesswork employees are currently doing on their own. Leaders who build that structure now will convert pilots into adoption. Leaders who skip it will keep mistaking usage for trust, right up until the gap the public is already living through shows up inside their own walls.
If your organization has AI tools in place but cannot point to measurable adoption, or you are the executive fielding the "is this actually working" question from your board, that is exactly the conversation worth having before the next license renewal.
Trust built for one careful rollout is a launch. Trust built into how every AI decision gets reviewed is a system.
Everything in the Gallup data points the same direction: transparency, independent review, and real training close the trust gap, while good intentions alone do not. Left to individual managers, that discipline drifts the moment a deadline gets tight, oversight becomes a rubber stamp, and training reverts to employees teaching themselves on the side.
Leadership Operating System
A trust gap this consistent across a national sample is a structural gap, not a discipline problem inside your team. See how a Leadership OS review works.
AI Adoption Advisory
If your AI tools have usage but no measurable trust or adoption, the fix is the governance and training structure behind the tool, not another license. See how fractional AI adoption advisory works.
Book a call Either track starts the same way.
Michael D. Levitt is the founder of Breakfast Leadership Network and the author of Burnout Proof and Workplace Culture. More writing at the Breakfast Leadership blog.
Additional Resources
Burnout In The Age Of AI: How Leaders Can Fix Broken Workloads, Not Just Add More Tools
Unleashing The Potential Of AI In Talent Acquisition: Redefining Recruitment Practices
Sources
Gallup and the Special Competitive Studies Project, "American AI Sentiment in 2026" (nationally representative survey of 3,152 U.S. adults, conducted June 2 to 23, 2026). All statistics in this article are drawn directly from this report.
