Almost everyone has tried AI, and almost no one is sure how they feel about it. It seems “grudging use” is the dominant sentiment, with countless other surveys of ours showing increasing adoption in tandem with recognition of AI’s limitations. It makes work more efficient– but adds to tool sprawl. It has good output– but untrustworthy CEOs. This tension is at the heart of our latest survey of U.S. consumers: AI has gone from novelty to default in record time, answering questions, planning trips, and optimizing workflow (in theory), yet it hasn’t seen mass enthusiasm similar to the smartphone or the internet before it.
We wanted to know what Americans actually think now that the hype cycle has had a few years to iterate. Do they believe AI is making their lives better, or just making them lazier? Do they trust the companies building it? Would they pay for any of it? And how does all of this fit into a broader, quietly souring relationship with the internet itself? The answers paint a picture of a country that uses AI constantly and believes in it conditionally.
Key takeaways: Americans can't quit AI, despite doubts
- 96% of Americans have tried AI, and 86% use it currently. Of those users, 98% find it useful and 86% say it improves their lives.
- 51% say AI is overhyped, 77% say it makes people lazier, and 57% say it weakens critical thinking.
- Only 26% of users currently pay for AI, and 54% say they would stop using it entirely if it required payment.
- Americans trust AI more than social media, government agencies, search engines, journalists, and employers; this says as much about those institutions as it does about AI.
- Given the choice between investing in AI and investing in manpower, 68% would rather invest in people.
How U.S. consumers feel about AI: 51% say it's overhyped, 54% won’t pay
AI has reached near-universal dabbling, if not adoption: 96% of Americans have tried AI, and 86% are active users today. For the people who use it, the day-to-day experience is overwhelmingly positive — 98% find AI useful, 86% say it improves their lives, and 75% say it makes them more productive.
So far, this aligns with the endless LinkedIn posts about B2B SaaS and AI solutions. But scratch the surface and the enthusiasm diminishes. A slim majority (51%) think AI is overhyped, more than 3 in 4 (77%) say it makes people lazier, and 57% believe it weakens critical thinking. In other words, Americans are using a tool every day that most of them suspect is quietly dulling the very skills that make them good at their jobs. Useful, yes— but at what cost?
Over half of consumers don’t think AI is worth paying for
That ambivalence shows up most clearly at the cash register. Despite near-universal use, only 18% of users say AI actually makes them money, and just 26% currently pay for it. What’s most telling and perhaps concerning for the AI companies on the verge of IPO is that 54% say they would stop using AI altogether if it came with a price tag.
Where AI does win is on trust, though it's a concerning victory. Americans trust AI more than several institutions that rely on public confidence:
- Social media: 34%
- Government agencies: 20%
- Search engines: 17%
- Journalists: 14%
- Employers: 8%
AI, in its ability to synthesize large swaths of the internet, has the credibility of the golden average– it’s as likely to be telling the truth as any average can. Unfortunately, ads and agendas have eroded trust in other institutions to the point that this logic of averages is a better bet.
American AI preferences: ChatGPT leads on trust, Grok lags, and 68% would rather invest in people
When Americans reach for AI, they overwhelmingly reach for it as a search tool. Here's how they actually use it:
- Searching for information: 90%
- Product comparison: 54%
- Planning: 51%
- Health and mental health advice: 42%
- Academic help: 29%
- Companionship or venting: 22%
That last figure is worth sitting with. More than 1 in 5 people are turning to AI for companionship or to vent— a quiet echo of a broader loneliness epidemic that's reshaping how Americans seek connection. And 42% are using AI for health and mental health advice, a high-stakes use case for a technology most users say they don't fully trust.
Search habits are clearly in flux, but traditional search still holds the line — for now. When asked how they prefer to look things up:
- 51% choose traditional Google search
- 25% choose Google's AI Mode / AI Overviews
- 15% choose a non-Google AI tool
- 9% choose a non-Google search engine
Add it up and roughly 4 in 10 Americans now prefer some form of AI-driven answer over the classic blue links. The "ten blue links" era isn't over, but it's no longer the default for half the country. This makes AI visibility all the more important (and we can help with that).
What AI platform do Americans trust the most?
Not all AI platforms are created equal in the public's eyes. When it comes to trust, there's a clear pecking order:
- ChatGPT: 42%
- Gemini / Google AI Mode: 21%
- Claude: 12%
And a clear doghouse. The least trusted results:
- Grok: 32%
- "All of them": 26%
- ChatGPT: 11%
ChatGPT's position is fraught. It's both the most trusted platform and one of the most distrusted, the inevitable lightning-rod status of being the category's default name and one of the first players in the field. Grok, meanwhile, lands firmly at the bottom. And the fact that "all of them" is the second-most-common answer for least trusted speaks volumes: for a quarter of Americans, the honest answer about which AI they trust least is simply "yes."
Skepticism extends to the mission statements the AI labs love to cite. 51% say OpenAI is not living up to its stated mission, and 38% say the same of Anthropic. The companies promising to build AI for the benefit of humanity have not yet convinced a majority of humanity (or at least U.S. consumers) that they're doing it.
Perhaps the most pointed finding of the whole survey: given the option between investing in AI and investing in manpower, 68% of Americans would rather invest in people. After years of being told that AI will replace human labor, more than 2 in 3 consumers are casting their vote the other way.
AI isn’t simply a tool; its infrastructural requirements have started hitting American wallets, particularly around the placement and powering of data centers. 93% say residents should have a choice about whether AI data centers get built in their communities, and 70% believe those data centers aren't worth the environmental impact. As the physical infrastructure behind AI becomes impossible to ignore, the public is signaling AI might not be worth all the drawbacks.
Bots on Bots: 40% believe in dead internet theory
Our browsing habits have narrowed accordingly. 70% of Americans visit 10 or fewer websites on a typical day, and 26% say they visit fewer sites than they did a decade ago. The sprawling, exploratory web of the early 2010s has collapsed into a handful of destinations — and for many people, 42% say they spend most of their internet time on social media, which is doing the bulk of the consolidating. Requiem for Geocities, indeed.
It's not making us feel good. 15% say they feel worse after using the internet, and 34% would pay for an ad-free version of it — a meaningful chunk of people ready to put money down just to make the experience less hostile. (Notably, that's higher than the 26% willing to pay for AI.)
Then there's the finding that ties the whole mood together: 40% of Americans believe in "dead internet theory" — the idea that the internet is now primarily bots talking to other bots, with real human activity drowned out by automated noise. Once a fringe conspiracy, it's now a belief held by 2 in 5 people. And as AI-generated content floods the web and the agentic era dawns, that suspicion is only going to get harder to dismiss. The irony is hard to miss: the same AI tools Americans use every day are, in their eyes, helping to hollow out the internet they've already fallen out of love with.
Ultimately, consumers have adopted AI completely and embraced it agnostically. They find it useful but suspect it's making them worse. They use it but (mostly) won't pay for it. They trust it more than their institutions but less than they trust their own instinct that people, not platforms, are the better investment.
At Howdy.com, we help American companies build world-class teams of human experts across Latin America — engineers, designers, and operators who know how to put AI to work without being replaced by it. If you'd rather invest in people than in hype, let's talk.
Methodology
In June 2026, we surveyed 1,008 Americans nationwide on how they feel about and use AI and the internet. Ages ranged from 18-65 with an average age of 39. 49% were women, 49% men, and 2% either nonbinary or chose not to disclose.
For media inquiries, reach out to media@digitalthirdcoast.net
Fair use
When citing this data, please attribute by linking directly to this page or Howdy.com.
