This summer I spoke with the CEO of Springboards, a startup building a large language model (LLM) intended to produce a wider variety of responses than mainstream competitors. Early in the call he said, “We often say that we’re a self-loathing AI company. We don’t know if we really like what we’re doing.” That remark captures the broader ambivalence around AI today: attraction to its capabilities paired with deep unease about how it is developed and deployed.
Public sentiment is turning sour in many places. The Pew Research Center finds that more US adults expect AI to have a negative personal and societal impact than expect positive outcomes, with the pessimism strongest among younger people. A report from Stanford University shows that more than half of people worldwide say AI products and services make them nervous. In a Gallup poll in May, 71% of US adults said they would oppose construction of a new AI data center in their area, compared with 53% who said they would oppose a new nuclear power plant. An NBC poll in March found AI to be less popular than ICE (Immigration and Customs Enforcement).
At the same time, AI adoption continues to surge. Market analyst Sensor Tower reported that ChatGPT reached one billion monthly users in May; Google DeepMind’s Gemini recorded about 950 million users in July. According to Pew, half of US adults now say they use a chatbot, more than twice the share reporting that in 2023, and one in four say they use one daily. Across the 38 OECD countries, more than a third of adults reported using generative AI tools in the past three months.
These figures describe two coexisting realities: a substantial number of people dislike or fear AI, yet many more are actively using it. It’s unlikely these are entirely separate populations — the overlap is probably significant. Usage can breed criticism: regions with higher adoption (the Global North) tend to be more pessimistic, while the Global South displays greater optimism. But a deeper explanation may be that people aren’t rejecting the technology itself so much as the relentless way major companies push AI into every domain and frame it as an epochal transformation.
This dynamic echoes the rise of social media over the past two decades. Billions adopted Facebook, Twitter and Google search even as criticism of the companies behind them mounted. With social platforms, opting out often meant losing content and connections; the cost of exit was high. With AI, however, there remains more potential for regulatory response and consumer choice. Political appetite to act appears stronger: all 50 US states have either passed or proposed laws to govern AI development and deployment, creating a patchwork of more than 2,100 bills nationwide — about a tenfold increase in three years.
There are also competitive alternatives emerging. Several high-quality open-source options exist that challenge Google, OpenAI and Anthropic, which could, for now, increase consumer choice and exert market pressure.
That said, trillion-dollar companies are hard to move. But the future is not as inevitable as some industry rhetoric suggests. When I asked the Springboards CEO why his company was building another model if he felt uneasy about AI, he said there was no walking back from LLMs — but it was possible to make them do different things.
What’s needed is more of that difference: clearer public communication about what AI can and cannot do, and technologies marketed without the apocalyptic sales pitch. If developers, regulators and users engage constructively, the next phase of AI could be shaped toward useful, explainable and less coercive applications — rather than an unstoppable tide decided by a few dominant players.
Why this matters
- Rapid adoption alongside rising anxiety will shape both regulation and market structure.
- With roughly one billion monthly ChatGPT users and thousands of AI-related bills under consideration, AI’s societal impact is likely to be long-lasting.
How these tensions are resolved—through law, competition, and product design—will determine whether AI remains a mixed blessing or becomes something more uniformly beneficial or harmful.
