The driver in the car next to me laughed at a headline about ChatGPT and then scrolled to a story about layoffs. I felt a small chill—the same tech that solves a spreadsheet can steal a paycheck. You and I are living inside that contradiction.
I read the new Bentley University–Gallup survey, which Gizmodo shared, so you don’t have to scan the charts on your phone between meetings. Let me walk you through what the data actually says, why the shift matters, and what companies like OpenAI, Google, Microsoft, Tesla and Waymo are getting wrong when they argue that more use equals more trust.
At a campus café I heard students call AI “useful but spooky” — Young adults are turning sharply negative
That anecdote tracks with the numbers: 47% of 18- to 29-year-olds now say AI does more harm than good, up from 27% in 2024 and 36% in 2025. In plain terms, the people closest to new tools are the most worried.
More people say they feel knowledgeable about AI—about 70% now versus 64% last year—and they increasingly believe tools like ChatGPT, Google Bard and GitHub Copilot can match human performance in tasks from homework help to basic financial guidance. Yet that growing familiarity is not breeding comfort. If anything, it’s sharpening doubts.
Do Americans trust AI?
Short answer: not much. Nearly three in four respondents said they trust companies to use AI responsibly “not much” or “not at all.” Among younger people, that distrust climbs to four in five. Even Republicans, who remain the most trusting group, dropped from 41% expressing some or a lot of trust in 2025 to about 35% now.
In my inbox PR teams still pitch “AI” as a badge of progress — But corporate trust is collapsing
Press releases trumpet partnerships between Microsoft and OpenAI, or chip announcements from NVIDIA, while customers grow skeptical of the people who sell them solutions.
Trust and marketing are diverging. Consumers increasingly treat corporate promises as noise—especially when they see layoffs, biased recommendations or opaque decision-making. You can say a model is “trained on diverse data” all you want, but when an app recommends a product or a diagnosis and people perceive harm, the PR line becomes meaningless.
Can AI replace human experts?
People think AI can perform as well as humans in narrow tasks—financial advice, homework support, basic medical triage—yet they still prefer human recommendations for driving decisions and complex medical guidance. Brands like Mayo Clinic or certified clinicians retain authority even as algorithms improve.
A colleague told me “I’m watching my role get automated” — Job fears are now widespread
That fear is statistical: 79% of respondents believe AI will reduce the total number of jobs—the highest level in four years of polling. Concern is strongest among middle-aged Americans; 84% of people aged 45–59 anticipate job reductions, up from 75% last year.
Tools such as GitHub Copilot and automated customer-service bots streamline work, and anyone paying attention knows efficiency can remove roles faster than markets create replacements. The anxiety you feel when a meeting mentions “AI integration” is not paranoia; it’s forecasting that has landed in public sentiment.
At a local DMV I watched a kiosk suggest a form—people liked the help but hated the outcome — The better AI gets, the worse it feels
Here’s the paradox: performance and trust are decoupling. The survey shows people increasingly see AI as comparable to or better than humans in specific tasks, while faith that it will improve lives or be used responsibly falls.
This is where metaphors help: AI is like a precision scalpel losing its edge—the tool is powerful but people fear the hand guiding it. It is also like a rising tide eroding a seawall—progress keeps coming, and institutions scramble to hold the line.
When companies promise faster service or cheaper advice without clear guardrails, users experience the harm first and the benefit second. That sequence matters. You trust systems that protect you before they help you, not the other way around.
Will AI take our jobs?
The short answer is nuanced: automation will change labor markets dramatically, displacing some roles and reshaping others. The survey suggests most Americans expect a net loss in jobs, and that expectation colors trust, politics and consumer behavior.
If you run product, policy or investment, hear this: improving accuracy or utility is not the same as building social license. OpenAI and Microsoft may ship better models; Tesla and Waymo may improve autonomy; but without transparent governance, responsible rollout and visible remedies for harms, the public will keep pulling back.
I don’t have a single fix, but I’ve watched how trust rebuilds: clear accountability, meaningful oversight, and real-world pilots that prioritize safety over headlines. If companies offered that, you might see a different curve in the next survey—but will they change fast enough to stop the skepticism from hardening into policy and market backlash?