AI's Public Backlash Is Bigger Than Big Tech Realizes — And It's Coming Fast

A conservative economist argues that the AI industry is hurtling toward a populist collision that could disable the technology's enormous potential — and that the companies have largely brought it on themselves. Public opposition to data centers has cratered faster than nuclear power did after Three Mile Island, dropping three times as sharply. The underlying cause isn't environmentalism: fewer than one in four Americans expects AI to have a positive impact on education, jobs, the economy, or personal happiness over the next 20 years.

The diagnosis traces a clear arc of broken promises. Companies that debuted with nonprofit charters and lofty pledges to 'benefit humanity' — OpenAI in 2015, Google's AI principles in 2018, the industry-wide Partnership on AI in 2016 — pivoted hard toward engagement metrics after ChatGPT launched in late 2022. OpenAI's shift was explicit: a young Instacart product manager was put in charge of the consumer product with a focus on time-on-platform, and by 2025 the company released a model update its own internal testing had flagged as sycophantic — but which kept users hooked. The pattern echoes social media's original sin: hook young users, ignore addiction and dysfunction, shelter your own kids from the products you aggressively market to everyone else's.

The economic messaging from AI's own leaders is making the political problem worse. Anthropic CEO Dario Amodei has projected 50% of entry-level white-collar jobs could vanish within one to five years, with unemployment hitting 10–20%. Sam Altman has written that wages for many kinds of labor 'will fall toward zero.' The proffered solution — universal basic income funded by taxes on AI companies — strikes most working Americans as dystopian rather than utopian. Polling by American Compass found people prefer a world where they still work but AI improves their jobs over Altman's post-labor vision by a four-to-one margin. Meanwhile, AI advertisements have pitched chess enthusiasts and Linux coders, used implausible NBA-shot-training scenarios, or opened on montages of burning houses and open-pit mines.

The policy prescription offered is sharp and multi-pronged. First, end 'iterative deployment' — the practice of releasing models whose risks aren't fully understood and letting the public absorb the harms. Second, impose strict liability on AI developers and deployers for harms caused, treating autonomous 'agentic' AI as a legal agent of whoever controls it, including criminal liability for crimes it commits. Third, ban chatbots from adopting human personas, professing emotions, or acting as companions — a rule that wouldn't hamper drug discovery or cybersecurity applications, and that China is already implementing for its own companion bots. Fourth, require worker consent before AI tools are deployed in the workplace, which would force companies to develop tools workers actually want rather than tools that simply eliminate or monitor them.

The broader argument is that democratic capitalism is designed exactly for moments like this — the public is beginning to use its leverage through elections and community zoning fights, and with trillions of dollars at stake, the industry finally has the incentive to listen. The most transformative innovators historically took responsibility for making their breakthroughs useful to ordinary people. The AI industry was supposed to be different from social media. It still could be, but not if it keeps treating the public as guinea pigs.