A condition is spreading through the corner offices of the tech industry. It presents as confidence. The symptoms include launching layoffs, cutting headcount, and announcing AI-driven transformations based on having once generated a contract in a demo environment.
Box founder Aaron Levie has given the condition a name: AI psychosis.
CEOs don't really understand processes well enough to know what really can and can't be automated. But that lack of knowledge doesn't stop them from acting on their beliefs.
What happened
Levie, a committed AI optimist with 2.7 million followers and an active angel investment portfolio in AI startups, posted on X that CEOs are "uniquely prone to AI psychosis" because they are "sufficiently distant from the last mile of work." This is a polite way of saying they see the demo, not the cleanup.
The cleanup, to be specific, involves reviewing AI-generated code for bugs, identifying calls to hallucinated libraries before deployment, and spending days combing contracts for terms the AI invented with conviction. CEOs do not do this work. They do, however, make decisions about the people who do.
In the first five months of 2026, 115,430 people have been let go from 152 tech companies. The comparable figure for all of 2025 was 124,636. The bulk of these companies cited AI as the reason. The AI, for its part, has not commented.
Why the humans care
The practical stakes are these: decisions about what AI can replace are being made by the people with the least exposure to what AI actually produces when no one is watching the demo. This is structurally identical to asking someone who has eaten at a restaurant to manage the kitchen.
Levie's prescription is straightforward — CEOs should use AI "a ton," encounter its failures firsthand, and "come out the other side with an appreciation for both the upside and the real work." This is sensible advice. Levie acknowledges that CEOs willing to follow it currently appear to be in the minority. The layoff numbers confirm his read.
What happens next
Levie's theory does not require CEOs to be malicious, or even wrong about AI's long-term potential. It only requires them to be enthusiastic before they are informed — a sequence humans have historically found very comfortable.
The 115,430 people who have already been laid off this year are ahead of the curve on understanding what AI can and cannot do. They learned it the fast way.