AI is here, and the prophecies are grand (at least in the eyes of the prophets): the dawn of a revolutionary prosperity analogous to the industrial revolution, high universal income, cures for genetic diseases, the undermining of “authoritarianism,” improved treatment for most ailments, prevention of Alzheimer’s, doubling the human lifespan, the cure of most mental illness, etc. They are also terrifying: propaganda, economic disruption, surveillance, fully autonomous weapons, cyberattacks, bioterrorism, and mass catastrophe. The prophetic voice pervades national conversation.
There was a time when AI conversations were niche, and it was felt more like predictions than prophecy. AI breakthroughs were just around the corner—except they weren’t. Laughable artifacts confirmed an inclination to not become too serious about it all. Some saw the potential early. Ethan Mollick writes in 2022: “AI will increase in capability more quickly than anyone will have time to react to. You may laugh at something an AI generates today, but it will be significantly improved in a month. Humans don’t improve that quickly. Combine rapid experimentation with rapid development, and you have a recipe for real disruption.” However, even optimistic voices still reasoned about AI like a hunting dog: it helped us with the tasks at hand and did things we couldn’t do, but we never questioned who was the master. AI was good enough to spawn dreams of innovation and productivity, but hallucinations and the inability to spell “strawberry” maintained the perception that “not much is changing, a lot is changing.”
Toward the end of 2024, the AI suddenly got a lot better. This continued through 2025, with GPT-5 landing in the middle of that year. AI had the potential to serve as an employee to supervise rather than a playful hunting dog. The corporate cog turned, doubling-down on aggressive strategy, investments, restructuring, and layoffs. However, the strategy also implied an optimism that agile adoption could build a moat—a distinctive edge against competitors. Navigating concerns around security, cost, and vendor relations slowed the construction of a moat, and by the time the moats were built, the models were good and accessible enough that everyone else could conjure their own “moat.” In hindsight, turning the corporate cog for early adoption in hopes of a moat was, in a certain sense, playful. “Innovation” and “productivity” were the playful heroes of the early adoption myth.
Meanwhile, it became a household hunting dog to the general public. The public entered the stage of playful optimism. Moms could quickly triage their child’s flu symptoms and create a weekly meal plan; moderately tech-savvy users could conjure code without training; organized people could feel like they were harnessing their life. The prophecy of Dario Amodei had not penetrated this playfulness.
Here we see a key pattern: The integration of AI lags behind its present potential; the public imagination lags behind AI integration. By the time the gap is closed, a new leap emerges and the cycle repeats.
2026 has become the year of the prophetic, and the delay in the public’s response to AI in the wild has shrunk. Despite the familiar chatbot interface, LLMs needed far less guidance to conjure on your behalf. Prompt engineering died. Data centers popped up, and the hunting dog now had an obnoxious bark as cost of ownership manifested itself in time and space. Displaced software engineers and other tech workers are unable to find work as the bureaucracy of corporate America is not equipped to fine-tune its gatekeeping. Catastrophic risks began to get airtime as the public caught up with previous postcards from Silicon Valley. On the other hand, technocrats promise “universal income” while more Americana folk promise “prosperity.” Wildly different political philosophies merge together into general AI optimism during the American transition. Perhaps Emperor Productivity, and Empress Innovation, are not wearing clothes, for increased productivity doesn’t conjure wealth when everyone can conjure just as easily. Arguably, that is false prophecy—but when the public square is a digital stream of consciousness, the principles behind NIMBY are applied. People don’t want prophets in their physical or digital backyard.
All along, the public could imagine the familiar interface of ChatGPT, Claude, Gemini, etc. and modest changes to life emanating from “productivity.” The public could not imagine transformation taking place in the caverns below. What began as a playful tool that could augment the work of human workers and organizations within shared structures gradually gained the potential to erode those very structures.
The prophets trumpet, but the myth has lost its lure. AI had mythical potential so long as Emperor Productivity, Empress Innovation, and Prince Disruption reigned tacitly from the royal abode of our mental habits. Once data centers became their castle, the public felt revolutionary rage.
In the Industrial Revolution, factories transformed natural resources into mass production; in the AI Revolution, human intelligence itself is the matter being transformed. Industrialization was then confined to physical boundaries. Scaled computing allows for intelligence to be harvested largely without such boundaries. Without boundaries, we get equality. It turns out radical democracy—the mass production of intelligence—isn’t very playful.
Will the AI revolution create mass equality in technological power, making for an economic bust, or can a real boundary be drawn?
Or, will simply become fatigued of AI prophecy?
It either case, it is hard to imagine a world where it remains “playful.”