The technology industry has always attracted visionaries who see around corners, but something darker has taken hold in the executive suites of Silicon Valley. Call it AI psychosis: a condition in which otherwise intelligent leaders abandon basic epistemic hygiene in favor of grandiose claims about artificial intelligence that bear little resemblance to technical reality.
The symptoms are now widespread enough to constitute an epidemic. CEOs announce products that do not exist, promise capabilities their systems cannot deliver, and speak of artificial general intelligence as though it were arriving next quarter. When pressed on timelines or technical specifics, they retreat into vague prophecy. The pattern has become so pronounced that investors, employees, and journalists have begun using clinical language—not as metaphor, but as the most accurate available description.
The pathology of the pivot
What distinguishes the current moment from previous hype cycles is the velocity of belief revision. Executives who spent years building businesses around specific value propositions have wholesale abandoned those identities to chase AI adjacency. Companies that made databases now make "AI infrastructure." Firms that sold advertising now sell "intelligence." The pivot is not merely strategic repositioning—it is a kind of personality dissolution, where the core identity of an organization becomes whatever investors seem to want to hear this week.
The financial incentives are obvious: AI-adjacent valuations have decoupled from revenue in ways that reward narrative over substance. But incentives alone cannot explain the fervor. Something about the technology itself—or perhaps the stories we tell about it—seems to disable the normal skepticism that executives apply to other domains. A CEO who would never claim to have cured cancer will casually announce that their chatbot understands human emotion.
The emperor's neural network
The most troubling aspect of AI psychosis is its social contagion. When one major CEO makes an outlandish claim, competitors feel pressure to match or exceed it. Boards, often lacking technical sophistication, cannot distinguish between plausible ambition and fantasy. Employees who raise concerns are labeled insufficiently visionary. The result is an information environment where the most extreme claims face the least resistance, while measured assessments are treated as evidence of inadequate commitment.
This dynamic has real consequences. Engineers are asked to deliver on impossible timelines. Products ship before they work. Users develop expectations that technology cannot meet, then lose trust when reality intrudes. The gap between promise and delivery is not merely embarrassing—it is corrosive to the broader credibility of technological progress.
Our take
The charitable interpretation is that these executives genuinely believe what they are saying, which is more alarming than the alternative. Cynical hype eventually exhausts itself; sincere delusion tends to escalate until external reality intervenes, usually painfully. The AI industry would benefit from a few prominent leaders willing to say what they do not know, what their systems cannot do, and which problems remain genuinely hard. Until then, the diagnosis stands: the people building our future have lost contact with the present.
