The AI Backlash Has Begun — And This Time, It’s Not Going Away

On a warm Sunday in May 2026, former Google CEO Eric Schmidt stood before the graduating class at the University of Arizona, adjusted his microphone, and delivered what he believed would be an inspiring line. “The question,” he said, “is whether you will have shaped artificial intelligence.” The response was not applause. It was boos. Loud, sustained, unmistakable boos — from the very generation of young people that the tech industry had assumed would be its most enthusiastic champions.

It wasn’t an isolated incident. The week before, a real estate executive was jeered at the University of Central Florida for calling AI “the next industrial revolution.” A record producer at Middle Tennessee State University received a similarly hostile reception. Three commencement ceremonies. Three pro-AI speeches. Three crowds that had heard enough.

Something has shifted. The AI boom that began in late 2022 with the launch of ChatGPT has, for three years, been treated as an unchallengeable force of nature — a wave so large and inevitable that resistance was not merely futile, but foolish. That assumption is now dead. The backlash is here. And unlike previous tech panics, this one has real teeth.

To understand why the backlash is so fierce, you have to understand how relentlessly AI has been sold. Every major tech company, every venture capitalist, every business publication, and every conference panel has spent three years telling anyone who would listen that AI would cure cancer, end poverty, make everyone more productive, and usher in a golden age of human creativity. The promises were extraordinary. The timeline was always “soon.” The evidence was frequently thin.

Meanwhile, the reality on the ground looked different. Companies were announcing mass layoffs and explicitly citing AI as the reason. Amazon, Microsoft, Salesforce, and IBM alone linked nearly 50,000 job cuts to AI-driven efficiencies in 2025. Early-career workers — the graduates being booed at — found that internships and entry-level positions were evaporating precisely when they needed them most. The same technology celebrated at TED Talks was quietly dismantling the career ladder beneath them.

The cruelest irony is that the people most hurt by AI’s rise are the same people most encouraged to celebrate it. Young graduates are told AI will create more jobs than it destroys — eventually — while watching their own opportunities shrink in real time. Workers are told to “upskill” while their employers simultaneously use AI to justify not hiring the newly upskilled. The cognitive dissonance has become impossible to ignore.

Backlash is easy to dismiss when it’s just angry tweets. What makes this moment different is that the resistance is showing up in places that actually matter: campuses, city councils, courtrooms, and churches. Here is the evidence that this is not noise.

None of these are fringe events. A papal encyclical shapes the views of 1.4 billion Catholics worldwide. Community resistance to data centres is now influencing elections in swing states. Open-source developer communities are the people who literally build the internet. When these groups turn against you, you have a serious problem.

For years, the environmental cost of AI was an asterisk — acknowledged in footnotes, minimised in press releases, and almost never discussed at the product launch events where AI was celebrated. That era is over. The numbers have become too large to hide.

Global data centre electricity consumption, which was already at 460 terawatt-hours in 2022, is now projected to exceed 1,000 TWh in 2026 — more than doubling in four years. Training a single large language model emits hundreds of tonnes of CO₂. Some hyperscale data centres consume up to 22 million litres of potable water per day for cooling. In regions already facing water stress — large parts of Africa, the American Southwest, and South Asia — this is not an abstract concern. It is a direct competition for survival resources.

The political heat is intensifying. US Senators Elizabeth Warren, Chris Van Hollen, and Richard Blumenthal publicly pressed major tech firms about AI data centres driving up residential electricity bills — a cost being passed directly to ordinary households who receive no share of the AI profits driving the consumption. President Trump, not typically aligned with progressive senators on anything, has also called for tech companies to pay their “fair share” of utility costs. When left and right agree that tech is extracting too much and giving back too little, a political reckoning is coming.

Let’s be fair. The pro-AI case is not without merit, and intellectual honesty demands we engage with it directly rather than dismiss it. AI has produced genuine breakthroughs in medicine, materials science, and climate modelling. It has made certain categories of knowledge work dramatically more productive. There are real and meaningful benefits.

It’s a reasonable argument. It may even be correct — eventually. But it carries a critical flaw that is becoming impossible to overlook: the timeline is vague, the benefits are concentrated, and the costs are immediate and widely distributed.

The people profiting from the AI boom are a small number of shareholders, executives, and investors in a handful of companies, most concentrated in a single American city. The people bearing the costs — job displacement, rising electricity bills, reduced device affordability, environmental degradation, and the psychological toll of technological uncertainty — are spread across every country on earth. The printing press did not send the profits of its revolution to fifteen people in Mainz while charging everyone else for the paper.

The argument that we should trust AI’s long-term benefits while ignoring its short-term costs is, at its core, asking the people who bear the costs to be patient while the people who capture the benefits get rich. That is not a stable social contract. Historically, it doesn’t hold.

Silicon Valley’s response to the backlash has been, so far, a masterclass in how not to handle public relations. When students boo your keynotes, the instinct of tech leaders has been to dismiss the audience as uninformed, emotional, or simply unable to grasp the magnitude of the opportunity. Eric Schmidt is “probably not too worried,” as one tech publication noted. That attitude is the problem.

The people booing are not Luddites. They are not afraid of technology. Many of them are studying computer science, data analytics, and — yes — artificial intelligence. They understand the technology perfectly well. What they are rejecting is the social arrangement being offered to them: a world in which AI generates extraordinary wealth for a small elite, disrupts their career prospects, drives up their utility bills, depletes their water supplies, and then asks them to be grateful for it.

The tech industry’s fundamental error has been to treat public acceptance as a communication problem rather than a distribution problem. The question is not how to better explain AI’s benefits. The question is how to ensure those benefits are actually shared. Those are entirely different challenges, and solving one does nothing to address the other.

The Samsung workers currently threatening to strike over their share of the AI memory boom are asking the same question as the graduating class of 2026. They made the chips that power the AI revolution. They want to know: where is our share? The industry has not provided a satisfying answer. Until it does, the boos will get louder.

Previous tech panics — about social media, about smartphones, about the internet itself — shared a common feature: they were largely cultural and psychological. People worried about attention spans, privacy, and social connection. Real concerns, but ones that could be managed, mitigated, or adapted to over time without fundamentally challenging the economic interests of a significant portion of the population.

The AI backlash is different because it is simultaneously cultural, economic, environmental, and political. It is being felt in pay-cheques, electricity bills, water supplies, and job markets. It is showing up in elections. It is drawing attention from the world’s largest religious institution. It spans every ideological spectrum — the left is concerned about labour displacement, the right about cultural change and energy costs, communities about local resources, developers about their tools, and young people about their futures.

A backlash that crosses every ideological line is not a backlash. It is a reckoning.

The AI industry has approximately one to two years, in my assessment, to demonstrate that its benefits can be distributed in ways that most people experience as fair — not just claimed in press releases, but genuinely felt in wages, energy prices, job availability, and environmental impact. If it fails to do so, the political and regulatory response will be severe, bipartisan, and global.