Americans are giving AI the side-eye. Donors should factor their feelings in before they begin giving to AI-related causes.
Americans aren’t confused about artificial intelligence. They increasingly use it more and understand it better, but they are unconvinced it’s good for them.
A new survey from Gallup and the Special Competitive Studies Project puts it plainly: “Americans' use of and familiarity with AI have grown, but their trust has not.”
That message should matter to donors asked to fund an AI literacy program, workforce initiative, or research into AI safety and oversight.
The numbers tell a clear story.
Reported AI use jumped from 39 percent of adults in 2025 to 51 percent in 2026, driven mostly by people over 30 catching up to a younger cohort that adopted early. Self-reported knowledge rose from 60 percent to 68 percent.
Trust in AI's ability to make fair decisions was flat at 28 percent. Nearly two-thirds of Americans say they distrust it outright. Familiarity hasn’t translated into confidence, which should give pause to funders whose theory of change assumes exposure alone will close the trust gap.
The AI economic mood has also soured.
In 2025, more than twice as many Americans believed AI would boost economic growth as believed it would hurt it. A year later, 45 percent expect AI to reduce growth, compared with 36 percent who expect it to help.
Seventy percent think AI will destroy more jobs than it creates, up roughly ten points in a year. Pessimism is sharpest among 18-to-29-year-olds, a group facing a marked experience gap, the distance between credentials they collect and the first rungs of a career ladder they need to climb.
Peter Cappelli, director of the Wharton School's Center for Human Resources, summarizes this dilemma: "Everybody wants somebody with three to five years' experience. Nobody wants to give those people right out of school any experience.”
Only 20 percent of Americans think most displaced workers will successfully retrain. That's a question of trust, not technique. Will community colleges, apprenticeship programs, local philanthropies, and the like assist people through disruption?
Despite these ominous rumblings, Americans' opinions converge on three issues.
First, 72 percent of Americans support expanding STEM education to prepare workers for AI in their future jobs, and 66 percent support employer-based training, both areas where support holds steady across party lines.
Second, 79 percent say government should prioritize safety and independent oversight even if that slows AI's rollout, also held across partisan lines since last year.
Third, 75 percent specifically want independent experts, not companies or government alone, to test AI systems before release.
Taken together, that's close to a bipartisan mandate on the core question of how AI should be developed. It also describes work philanthropy can do—significantly, it can fund independent capacity that neither industry nor government has an incentive to build on its own. This concern isn’t confined to jobs and fairness.
A Politico poll found that 63 percent of Americans believe advanced AI could eventually pose a serious risk to humanity, with 48 percent favoring a pause in developing more advanced models.
The finding also held up across party lines. Majorities of both Harris and Trump voters registered at least moderate concern, even as the size of that majority differed.
When people are this worried about a technology at the practical level of jobs and fairness, they’re not asking foundations to settle the argument. They want independent verification and transparency that the Gallup data already points toward.
Foundations don't need to referee whether AI will help or harm the economy. They need a more focused set of questions as they consider what to do, starting with these six.
- Will this program build independent capacity or amplify a company's claims about its tools? Transparency from companies, independent testing, and pre-deployment fairness checks are the three interventions Americans say would most increase their trust. A grant funding third-party testing answers a demand the public has named. A grant funding a company's own safety marketing doesn’t.
- Is the training aimed at the workers most exposed to disruption or those who have adapted? Older adults drove nearly all the year-over-year growth in AI use, while younger adults, already the heaviest users, grew more pessimistic about their prospects. Dollars may do more good reaching mid-career workers only now starting to engage than reinforcing the most AI-literate cohort.
- What happens after worker displacement? Only one in five Americans believes retraining mostly succeeds. Programs promising it should show a track record, not just a curriculum.
- Is this a K–12 or a workforce initiative? Americans are split, and more strongly opposed than supportive, on teaching AI skills inside core K-12 subjects. But they strongly back STEM investment and on-the-job training. A donor drawn to AI education should ask which area the grant actually serves.
- Would this grant exist, or look different, if trust in AI were already high? Trust has stayed flat near 28 percent even as use and familiarity have grown. That suggests the gap a program should close is about verification and confidence, not exposure. A grant that would look the same in a high-trust environment may be solving a different problem than the one Americans are describing.
- Is the foundation's own region facing a data center or infrastructure siting decision? If so, it should support an effort to bring independent, trusted information to that debate. Roughly six in 10 voters, across party lines, support limits on new AI data centers, driven by concern over local water supply and electricity prices. Many regional and community foundations can help inform that discussion.
None of this requires philanthropy to resolve whether AI will ultimately help or hurt American workers. The public has already decided that argument doesn’t need resolving before institutions act. What Americans want, across party lines, is training they can use now and oversight they can trust later.
Philanthropy has funded both before, in other transitions. This one requires applying that playbook with some urgency. A donor who funds toward that specific, well-documented demand is not guessing at public sentiment.
Bruno V. Manno is a senior advisor at the Progressive Policy Institute and a former U.S. Assistant Secretary of Education for Policy. He can be reached on LinkedIn.
