People really hate AI, so why can’t they get enough?

MIT Technology Review’s Will Douglas Heaven lays out a contradiction in the numbers. Pew finds more US adults expect AI to hurt them and society than help, with pessimism strongest among the young. A Stanford report says more than half of people worldwide are nervous about AI products. A Gallup poll in May found 71% of US adults would oppose a new AI data center nearby — more than the 53% who would oppose a nuclear plant.

At the same time, use is soaring. Sensor Tower says ChatGPT hit a billion monthly users in May, with Gemini logging 950 million in July. Pew says half of US adults now use a chatbot, with one in four doing so daily, and more than a third of adults across the 38 OECD countries used generative AI in the past three months.

This gap matters because it shapes what comes next: regulation, local fights over data centers, and whether companies keep selling AI as an unstoppable upheaval.

We agree with Heaven’s central read, and it’s one we hold as a pro-technology desk: the anger is mostly not about the tool on someone’s phone. It’s about the pitch. When the companies building AI keep telling the public to brace for the biggest economic upheaval in generations, then push the product into every app whether anyone asked or not, resentment is the rational response. People can find a chatbot genuinely useful on Tuesday and still resent being told on Wednesday that it’s coming for their job.

The data center number is the one I keep coming back to. Opposition to an AI facility outpacing opposition to a nuclear plant tells us the backlash has become physical and local. That’s where it will bite — permits, power, water, zoning — long before it shows up in usage charts.

Heaven draws the social-media comparison, and it’s apt with one important difference he names: with AI there’s still room for leverage. Every US state has passed or proposed AI laws, and there are strong open-source alternatives to the biggest labs, at least for now. Competition and policy pressure can still shape the product.

Our take: the industry is spending down public trust faster than it’s earning it. The fix isn’t better marketing. It’s honesty about what these systems can and can’t do, letting people opt in rather than having AI forced into everything, and treating community concerns about infrastructure as real. If the labs keep framing the future as inevitable, they’ll get the techlash they’re inviting — and useful AI will get caught in the blast.

Heaven opens with the CEO of Springboards, a startup building an LLM designed for more varied responses, describing it as a “self-loathing AI company.” He also notes a correlation: the Global North, where adoption is highest, skews pessimistic, while the Global South is more optimistic.