A widening chasm has opened between how Americans experience artificial intelligence. For a select group of developers and researchers, AI represents a fundamental shift in capability: software that writes itself, research problems solved autonomously, coding work handled by machines operating with minimal human guidance. For everyone else, it feels like window dressing on existing tools, a marginally smarter search engine or a faster email sorter.
This gap matters because trillions of dollars and millions of jobs hinge on a technology most Americans neither understand nor trust. The stakes are enormous, yet the opportunity to participate in AI's frontier remains locked behind walls of access, expertise and cost.
The clearest sign of this divide is the emergence of competing elite AI models designed almost exclusively for power users. OpenAI's Sol and Anthropic's Fable have become lightning rods for developers obsessed with frontier capabilities, each model optimized to run lengthy coding and research tasks with minimal intervention. This week, prominent AI researchers have spent time comparing the two systems like sports fans debating rival athletes, dissecting their temperaments and work styles with almost religious fervor.
One researcher tweeted that Fable operates like a "wise owl" thoughtfully working through problems, while describing Sol as "a rottweiler who will grab the problem by the throat and not let go until it is done." To frontier users, these distinctions matter deeply. To almost everyone else, they might as well be comparing spacecraft in a galaxy far away.
The numbers tell the story plainly. OpenAI reports over 900 million weekly ChatGPT users and more than 50 million paying subscribers. But the actual user base for cutting-edge agentic coding tools, the frontier technology that excites Silicon Valley, represents a fraction of a fraction. Nearly half of American adults use AI chatbots, yet the overwhelming majority use them for basic information retrieval, the same task Google has performed for two decades. Nobody is building companies on that.
Even access itself has become a status marker. Sol launched as a restricted preview for OpenAI's chosen partners before wider release. Fable was pulled offline globally for three weeks in June due to U.S. export controls, while its more powerful cousin, Mythos, remains locked behind a small circle of trusted organizations. Inside the tech world, the new pecking order is clear: free users at the bottom, then paid users, power users, preview access holders, and finally an insider class testing capabilities the broader world will never see.
The parallel to earlier technological divides is instructive. A century ago, electricity created a stark chasm between urban and rural America. By 1930, nearly nine in ten urban homes had power while only about one in ten farms did. Private utilities saw no profit in wiring scattered rural customers. Only a federally funded initiative, the Rural Electrification Administration, eventually bridged the gap through years of loans and infrastructure investment.
AI's divide may prove harder to close. Frontier access is inherently scarce and expensive. Even where free access exists, most people lack the technical foundation to use sophisticated tools productively. The industry is counting on this transformation becoming inevitable through sheer momentum, but legitimacy matters too. And legitimacy is eroding.
Public trust in AI has collapsed even as adoption has climbed. Sixty-three percent of Americans believe AI is advancing too quickly, and just 16 percent expect it to benefit society over the next 20 years, according to Pew Research. The clearest economic winners so far have been investors, technology companies and power users. Ordinary Americans face the disruption: job displacement, surging energy consumption from data centers, information feeds flooded with AI-generated content, all without clear upside for themselves.
The Trump administration published an AI literacy framework in February to help workers "share in the prosperity that AI will create." OpenAI, Anthropic, Microsoft and Amazon jointly committed $500 million in June to a workforce retraining program. But training programs have limits. Frontier users maintain advantages that persist regardless of policy: better tools, earlier access, deeper technical knowledge and hundreds of hours spent experimenting with systems that evolve every few weeks.
The AI industry is betting that abundance eventually flows downward, that revolutionary technology ultimately lifts all boats. History suggests they should worry. Technology needs permission from the society it transforms, not just inevitability. Right now, they are losing that permission.
Author James Rodriguez: "The AI industry wants to remake the world before most people understand what it does, and that's a recipe for backlash that could slow real innovation."
Comments