The most alarming thing about the current AI boom is not the technology itself but the people building it. A pattern has emerged among tech executives that psychiatrists might recognize: grandiose delusions, magical thinking, and an inability to distinguish between aspiration and achievement. Call it AI psychosis.
The condition manifests predictably. CEOs announce that artificial general intelligence is imminent, that their chatbots exhibit "sparks of consciousness," that humanity stands at the threshold of a new epoch. Meanwhile, their products hallucinate facts, struggle with basic arithmetic, and confidently generate nonsense that sounds plausible to humans who do not know better. The gap between rhetoric and reality has never been wider.
The feedback loop of delusion
What makes AI psychosis so persistent is the financial incentive structure that rewards it. Venture capitalists have poured hundreds of billions into AI startups over the past three years, and that capital flows most readily to founders who promise transformation rather than incremental improvement. Modesty is punished; hyperbole is rewarded with billion-dollar valuations.
The result is a selection effect. Executives who maintain epistemic humility—who acknowledge that large language models are sophisticated pattern-matchers rather than nascent minds—find themselves outcompeted for attention and funding by peers willing to invoke science fiction. The industry's discourse has drifted from engineering into eschatology.
The real casualties
The psychosis would be merely amusing if it did not have consequences. Companies are deploying AI systems in healthcare, criminal justice, and education based on capabilities that exist primarily in press releases. Workers are being displaced not because AI can actually do their jobs but because executives believe it can, or will soon. Regulatory frameworks are being shaped by policymakers who take Silicon Valley's self-assessments at face value.
Perhaps most troubling is the effect on the engineers themselves. Many of the researchers building these systems express private skepticism about their employers' public claims, yet the culture of AI companies increasingly demands fealty to the transformative narrative. Dissent is career-limiting.
Our take
The technology industry has always been prone to hype cycles, but AI psychosis represents something qualitatively different: a collective break with empirical reality by people who control enormous resources and influence. The cure is not more regulation or less investment but something simpler and harder—executives who can distinguish between what their products do and what they wish their products did. Until that happens, the rest of us should treat Silicon Valley's pronouncements about AI with the same skepticism we reserve for any other patient describing their symptoms.
