The AI everyone says they hate: why public rejection isn't slowing its mass adoption

The article, written by Will Douglas Heaven in MIT Technology Review, opens with a striking confession: the chief executive of Springboards —a startup developing an LLM designed to generate more varied responses than its rivals— described his own company as "a self-hating AI company," admitting they are not even sure whether they like what they do. The author picks up the thread and confesses to feeling something similar as a journalist covering AI: he loves his job, but not necessarily the technology he covers, which he sees as distorted by hype, hijacked by zealots and so omnipresent as to be inescapable.

That ambivalence, the article argues, is not an isolated case but the prevailing mood worldwide toward artificial intelligence: a deeply marked love-hate relationship. On one hand, public sentiment is souring fast. According to the Pew Research Center, more American adults believe AI will have a negative impact —both personally and for society at large— than expect a positive one, and the pessimism is strongest among young people. A report from Stanford University notes that more than half of the world's population says AI products and services make them nervous. A Gallup survey in May found that 71% of American adults would oppose the construction of a new AI data center in their area, a figure far above the 53% who would reject a new nuclear plant. And an NBC poll in March placed AI as less popular than ICE (the US immigration and customs agency).

On the other hand, the number of AI users keeps soaring. ChatGPT reached one billion monthly users in May, according to market analysis firm Sensor Tower, and Google DeepMind's Gemini was close behind with 950 million users in July. According to Pew, half of American adults say they already use a chatbot, more than double the figure in 2023, and one in four uses one daily. It is also a global habit: more than a third of adults across the 38 OECD countries (the group of the world's wealthiest democracies) say they have used generative AI tools in the past three months.

All these surveys, the author notes, point to two contradictory truths: many people can't stand AI and, at the same time, many people can't do without it. Faced with this paradox, he offers several hypotheses. It could be two distinct groups of people, but he doubts that: the numbers only add up if that Venn diagram tends to become a circle — that is, if to a large extent the same people who reject AI in surveys are also using it every day. He also considers that the more AI is used, the more negative opinion of it becomes, an idea supported by a certain correlation: the Global North, where adoption is highest, tends toward pessimism, while in the Global South, with lower adoption levels, optimism is greater.

But the explanation the author defends most strongly is a different one: when people say they hate AI, what they reject is not the technology itself but the relentless push by the companies behind it to force it into as many areas of everyday life as possible. Those companies warn society to brace for the biggest social and economic disruption in generations; it is only logical, the author reasons, that people do not take that message well.

The piece draws a parallel with what happened with social media over the past twenty years: billions of people joined Facebook and Twitter despite the growing "techlash" (public backlash against Big Tech) against the companies running them, and something similar happened with Google search. In that case, users had little room to maneuver: leaving a platform meant losing all the content and connections built up there, so the only real option was to put up with it or start from scratch somewhere else.

With AI, however, the author sees room for the public to have more ability to shape the technology's direction. On one hand, there is greater political appetite for regulation than there was in the social media era: all 50 US states have passed or proposed laws regulating AI development and deployment, creating a patchwork of more than 2,100 bills across the country, ten times as many as three years ago. On the other, top-tier open-source alternatives to Google, OpenAI and Anthropic already exist on the market, which opens the door —at least for now— to more consumer choice and the market pressure that comes with it.

The author is careful not to lapse into naive optimism: he acknowledges that trillion-dollar companies are hard to move. Even so, he insists that AI's future is not as written as those same companies tend to imply. When he asked the Springboards CEO why he kept building a new model if deep down he was not convinced by AI, the answer was that there is no going back on large language models, but that they can be made to do something different. The article closes with that wish: to see more different proposals, more honest clarity about what AI can and cannot do, and a technology that stops being sold as if it were about to take over the world.