Why AI Executives Still Don’t Understand the Public Backlash

Public skepticism about AI is growing faster than Silicon Valley seems willing to admit. A Pew Research Center survey published this week found that more than half of Americans under 30 are more concerned than excited about AI, a share that has grown 24 percent in five years. More than 70 percent of all age groups believe the technology will cost jobs. Yet recent public statements from Meta CEO Mark Zuckerberg and Anthropic CEO Dario Amodei suggest both are still diagnosing the wrong problem.

What happened

Data point	Detail
Americans under 30 more concerned than excited about AI	More than 50%, up 24% over five years (Pew Research, this week)
All age groups who think AI will reduce jobs	More than 70%
Zuckerberg AI essay length	6,500 words, published earlier this month
Effect of different AI messaging on public opinion	Little to no difference (Searchlight Institute survey)

The numbers from Pew are striking on their own. But what makes this moment notable is the public sparring between two of the industry’s most prominent figures over who is to blame for the mood. Zuckerberg took a shot at Amodei in his essay, suggesting that doom-laden messaging from AI developers was feeding public fear. Tech investor Gavin Baker echoed the point on social media, arguing that Amodei’s communication had directly helped efforts to block data center construction in the US. White House AI adviser David Sacks added that Anthropic’s public statements had “done more than anything to shape public fear.”

Amodei pushed back in a lengthy social media post. He called it “fundamentally a crisis of trust,” writing that ordinary people do not trust companies or governments and see AI as “the latest iteration” of tech finding new ways to exploit them. That framing is closer to what the data actually shows.

Where both executives lose the thread

Amodei’s trust argument holds up well until he pivots to arguing that the core criticism of AI companies is that they have not yet delivered on big promises, citing the oft-repeated goal of AI finding a cure for cancer. That framing still centers the problem on output rather than on how the technology is already reshaping daily life.

Zuckerberg’s essay describes a future where everyone has a personalized AI assistant with intimate knowledge of the user (drawn, he implies, from data Meta’s apps have already collected) helping with careers, hobbies, and even child-rearing. That product does not exist yet. And according to current polling, the people who would need to adopt it do not trust the company building it.

A survey from the Searchlight Institute makes the messaging-as-solution argument look even shakier: exposure to different AI narratives made little difference in how people assessed the technology’s impact on society. The backlash is not a branding problem.

What is actually driving the concern?

Polling points to anxieties well beyond job displacement. Americans also worry that AI will harm young people’s ability to form real relationships, damage students’ critical thinking, and allow tech companies to profit from creators’ work without compensation. Data centers have become a physical focal point for these frustrations. Tech companies have made commitments around water use, clean energy, and local investment, but those efforts have not moved public sentiment. As Wired’s Maxwell Zeff notes, the data center buildout has become “the physical manifestation of people’s hatred of AI and the ways tech companies engage with communities overall.”

Meanwhile, the rest of culture is capitalizing on the hostility. Beverage brand Liquid Death teamed up with Jason Kelce (former NFL player and Taylor Swift’s brother-in-law) on a commercial inviting people to urinate in bottles and mail them to AI data centers. Liquid Death’s creative lead told Marketing Dive that data center anger is “literally the one thing that unites all Americans right now.” That is a joke, but it is a joke that lands because it is grounded in something real.

Our take

From where we sit, the executives are making a classic attribution error: they are asking “why do people misunderstand us?” when the better question is “what have we actually done to earn trust?” Framing the backlash as a comms failure or a side-effect of scary CEO rhetoric puts the fix in the wrong place.

For businesses evaluating AI integration, this matters practically. Adoption inside organizations runs into the same trust wall. Employees pushed to use AI tools they distrust produce worse outcomes, not better ones. The companies that get adoption right tend to be the ones that start small, demonstrate value on a real task, and let people opt in rather than mandating the technology from the top. That is a slower path, but the data suggests it is the only one that actually works.

We cover how the broader AI landscape is shifting week to week in our AI news coverage. The public sentiment story is one worth watching closely, because it will shape which AI products businesses can actually roll out to customers without friction.

What to do about it

1. Audit which AI tools you are already using that touch customer-facing workflows, and check whether your customers know about them.
2. Give staff a clear, honest explanation of what data any AI tool sees and does not see before rolling it out.
3. Start with internal productivity use cases where trust stakes are lower, and build outward from demonstrated wins.
4. Watch adoption metrics, not just output metrics. A tool nobody uses is not saving you anything.

The companies that will win on AI adoption are the ones that treat trust as a product requirement, not an afterthought.

Source: WIRED · AI