“AI psychosis” in the executive suite: When managers massively overestimate the capabilities of AI

A prominent voice from the tech industry is speaking out in the debate about the actual potential of artificial intelligence in everyday business life. Aaron Levie, head of the cloud service Box, sees many managers having a distorted perception of the implementation of AI systems.

He was responding to a discussion on Platform X. The thesis was discussed there as to whether CEOs are particularly susceptible to illusions when it comes to AI. According to Levie, the problem lies in the distance of the management level from the concrete work steps that are essential for real added value in practice.

Simply looking at the end result clouds the view

The term AI or chatbot psychosis actually comes from medicine. He informally describes the emergence of clinical delusions through the use of language models.

Levie, on the other hand, uses the word metaphorically for managers who lose touch with reality because they usually only see embellished end results and error-free product prototypes.

If contracts are generated in a matter of seconds or impressive presentations are presented on the board, the impression is created that the technology is already functioning smoothly. The often tedious detailed work that is necessary to integrate AI models stably and error-free into an existing infrastructure usually remains hidden from those responsible.

Increased use as a medicine

In order to counteract this misperception, the boxing boss is calling for a rethink on the executive floor. In his opinion, the solution is not to distance oneself from technology. Unsurprisingly for a tech CEO, he instead recommends that managers use AI much more intensively in their own everyday lives.

Through personal use, they could develop a better understanding of how AI agents operate meaningfully in the company. Only through “trial and error” experience could the limits of the systems be explored. This creates the necessary balance between enthusiasm for the theoretical potential and respect for the actual hurdles.

This article was originally published on May 29, 2026, but is still of interest to many of our readers. That’s why we’ve updated it and made it available again here.