AI companies expected widespread use to make their technology feel ordinary and eventually welcome. Recent polling and local resistance point in the opposite direction: exposure has increased, but confidence has not followed. What once looked like a communications challenge is becoming a commercial and political constraint.
TechCrunch draws together several signals of that shift. A Pew Research study found 52% of Americans were more concerned than excited about AI’s growing role in daily life, up from 37% in 2021. A separate Economist/YouGov poll found more than 70% believed AI was advancing too quickly.
Adoption has not converted into confidence
The industry’s early assumption was straightforward. As AI moved into search, email, televisions and other familiar products, consumers would experience its benefits and resistance would fade. Instead, many people encounter features they did not request while hearing warnings about job losses, school cheating and the use of copyrighted work in model training.
A CNBC poll cited by TechCrunch asked adults aged 18 to 34 about nine prominent AI leaders. A majority did not trust those executives to act responsibly. Polls vary in wording and methodology, so they should not be read as a precise forecast of behaviour. Together, however, they indicate that scepticism extends beyond one company or one controversial product.
The cultural response may also be visible in interest in analogue or low-tech alternatives. TechCrunch points to renewed enthusiasm for dumbphones, point-and-shoot cameras, tape decks, classic iPods and in-person activities. Those trends have many possible causes, and the source does not prove that AI drove them. They nonetheless fit a desire for experiences perceived as more human, controllable or disconnected from algorithms.
Infrastructure makes the costs physically visible
Public concern becomes harder for technology companies to treat as abstract when data centers arrive in local communities. The National Republican Senatorial Committee reportedly warned major AI companies that opposition to facilities could hurt the party in an Ohio election. Meanwhile, developers are offering job guarantees, clean-water investments and other benefits to secure local support.
One Louisiana parish was reportedly offered $50,000 bonuses for teachers. Such concessions suggest that the cost of AI infrastructure now includes community acceptance alongside land, electricity, cooling and construction. Residents may be asked to tolerate water use, grid pressure or landscape changes before they see a direct improvement in their own lives.
That imbalance helps explain why technical progress alone may not build legitimacy. A more capable model is meaningful to developers and investors, but it does not automatically answer a household’s concerns about employment, education or utility prices. When promised benefits remain distant while infrastructure costs are immediate, distrust can affect permits, schedules and project economics.
Better messaging cannot substitute for useful outcomes
Some executives frame the backlash as a narrative problem. Airbnb chief executive Brian Chesky offered a broader diagnosis: the industry also needs products that ordinary people genuinely value. Anthropic chief executive Dario Amodei similarly described negative perception as a crisis of trust and acknowledged that AI companies have not yet delivered their largest promises to benefit society.
Those statements are executive assessments, not evidence that proposed breakthroughs such as curing cancer are close. They are notable because they shift responsibility away from an uninformed public. If people understand the trade-offs and remain unconvinced, another marketing campaign is unlikely to resolve the issue.
Trust can be rebuilt only through experience: products that solve recognisable problems, safeguards that work in practice and infrastructure agreements that distribute benefits credibly. The polling does not show that Americans reject every use of AI, nor does it predict permanent opposition. It does show that ubiquity should not be mistaken for consent.
For AI companies, that distinction now has measurable consequences. Public confidence influences customer adoption, political scrutiny and where the physical machinery of the industry can be built. Trust is no longer a soft reputation metric; it is becoming part of the operating environment.
