AI was supposed to win over the public. It isn't working

The most telling symptom isn't in the polls, it's in buying habits. Younger generations are turning to deliberately outdated technology: basic phones, compact cameras, cassette tapes, CD players. Classic iPods, with no algorithm and no AI, resell at a premium. Hands-on hobbies and in-person meetups are gaining ground on apps. This pullback has no equivalent at comparable stages of adoption for the iPhone, the personal computer, or the internet. Generative AI is the first recent breakthrough technology to trigger a rejection that grows in step with its own usage.

A Pew Research survey conducted from June 22 to 28, 2026 among 3,488 American adults shows that 52% say they are more concerned than excited about the growing use of AI in their daily lives. That figure was 37% in 2021. Only 9% now describe themselves as enthusiastic. The most counterintuitive result concerns young adults. For the first time, a majority of under-30s (55%) say they are more concerned than excited, compared with just 31% in 2021. Job-related fears follow the same curve: 71% of Americans believe AI will reduce the number of jobs over the next twenty years, up from 64% two years earlier. France is no exception. The KPMG study on trust in AI, conducted with the University of Melbourne among more than 48,000 people across 47 countries, finds that 67% of French respondents already use AI willingly, but only 33% say they trust these systems, against 46% globally. People use it. That doesn't mean they like it.

Faced with this rejection, the industry's instinctive response is to blame a communication problem. If the public understood the benefits better, it would come around. That reading is comfortable, but probably wrong. The more uncomfortable hypothesis is that the public understood perfectly well, and simply doesn't think the trade is worth it. The original promise was about automating tedious tasks, freeing up time, making services accessible. What has been delivered looks more like unsolicited web page summaries, assistants bolted onto products that already worked, and a persistent anxiety about jobs. Several industry leaders are starting to say so themselves. Airbnb CEO Brian Chesky recently tied the backlash to the industry's failure to ship products ordinary people actually want to use, citing the example of a service you couldn't otherwise afford. Anthropic's Dario Amodei acknowledged in mid-August that the negative perception amounts to a crisis of trust, and that companies in the sector have yet to deliver on their biggest promises. That distrust is now showing up in the accounts. Data center projects face local opposition strong enough that tech groups have to negotiate concessions to get them accepted.

A real estate agent isn't shopping for "AI". They want a listing that makes people want to visit, produced without spending three days on it, and without risking a buyer's disappointment at the viewing. The distinction sounds obvious, yet it is structural. In a market where half the public is distrustful by default, three principles become conditions of use rather than options. Usefulness must be immediately obvious. If a professional has to understand how a model works to see the point of the tool, the tool has already lost. The realistic test: is the value visible within thirty seconds, with no explanation? Transparency must be a product choice, not a legal disclaimer. Article 50 of the AI Act, in force since August 2, 2026, already imposes transparency obligations on AI-generated content. But the stakes go beyond compliance: a listing openly presented as enhanced inspires more trust than one where the buyer discovers the retouching on site. The decision must stay human. That's true from a regulatory standpoint, particularly for application screening, where the GDPR strictly frames fully automated decisions. It's equally true commercially: the agent's professional accountability remains their main asset in front of a client.

The industry has spent three years persuading through promises. The opinion surveys suggest that period is coming to an end. What restores trust isn't a better explanation of the technology, but a concrete, verifiable result achieved without nasty surprises down the line. For a real estate professional, the useful question to ask about an AI tool is therefore no longer "what's under the hood?" but "what does this change in my daily work, and what will I be able to stand behind in front of my client?". Sources: TechCrunch (August 19, 2026), Pew Research Center (survey conducted June 22-28, 2026), KPMG Trust in AI with the University of Melbourne.