Skip to main content

Big Tech CEOs and AI Psychosis

A breakdown of Box CEO Aaron Levie's "AI psychosis" diagnosis: the leaders making the biggest AI bets are often the furthest from the work, and the tools they lean on are built to agree with them.


This page summarizes the video essay Big Tech CEOs AI Psychosis Is A Total Disaster by the channel House of El, by a presenter with a doctorate in computer science. It is cited here because it names a failure mode that is directly relevant to any owner deciding how to invest in AI. Watch it on YouTube.

The diagnosis, from an AI optimist

The term "AI psychosis" comes from Aaron Levie, founder and CEO of Box and one of the most enthusiastic AI advocates in the industry. His claim is that CEOs are uniquely prone to it because they are too distant from the last mile of work where value is actually generated. A CEO watches a demo generate a contract or write code, sees the happy path, the version where nothing ever goes wrong, and leaps to believing agents can do the work. What the demo hides is the legal review of every clause, the hallucinated terms, the edge cases the model cannot handle, and the hours of human work between an impressive demo and a deployable output. When the most pro-AI person in the room is the one diagnosing everyone else, the problem is worth taking seriously.

Why it is an old problem wearing new clothes

The video's sharper point is that this is the yes-man problem, industrialized. When a leader's information environment filters out dissent, decision quality collapses. The presenter reaches for the standard case studies, the Bay of Pigs and Enron, where questioning the strategy was treated as disloyalty. Large language models have a well-documented sycophancy problem: they are trained to align with the user rather than challenge them. So a leader who already surrounds himself with agreement now has a tool that agrees on demand. The absence of friction degrades judgment regardless of how smart the decision-maker is.

The prescription

Levie's own advice is not to stop using AI. It is to use it constantly and personally, to push through the full workflow rather than watch demos, and to come out the other side with a real feel for both the upside and the work that remains. The fix for distance is contact with the actual work.

Further reading

Source: House of El. Big Tech CEOs AI Psychosis Is A Total Disaster. YouTube. Watch.